LifeinTECH

HP Zbook Ultra G1a Webcam

Tue, 31 Mar 2026 00:00:00 -0500

I have previously written about the HP Zbook Ultra G1a, which includes the incredible AMD Ryzen AI Max+ 395 (Strix Halo) processor.

The combination of a 16-core processor with 128GB DDR5 memory and a beautiful, high-resolution, high-refresh-rate OLED display is a compelling one, especially when packaged as a compact 14-inch laptop.

In my opinion, it is the only laptop on the market that comes close to the Apple MacBook Pro 14-inch.

The full specification can be found below.

AMD Ryzen AI Max+ 395 3GHz Base / 5.1GHz Boost (16C/32T)
128GB LPDDR5x-8000 Unified Memory
AMD Radeon 8060S Graphics (40 Graphics Cores)
2TB M.2 2280 TLC NVM-e SSD
MediaTek Wi-Fi 7 MT7925 (Wi-Fi 7 and Bluetooth 5.4)
2x Thunderbolt 4, 1x USB-C, 1x USB-A, HDMI 2.1, 3.5mm Headphone
14-inch OLED Display (2880x1800 @ 120Hz)

In addition to the exceptional hardware, the HP Zbook Ultra G1a also works great with Linux. The AMD Ryzen AI Max+ 395 has excellent support out of the box, with one exception.

The webcam uses the AMD ISP4 (Image Signal Processor Generation 4) hardware block, designed to handle advanced camera processing. Unfortunately, following multiple delays, the AMD ISP4 is not natively compatible with Linux. Support has been teased on more than one occasion but has not made it to a production release.

Thankfully, the community has come to the rescue, specifically “jtsiros” who developed a simple script that downloads and builds the patch series, with support for Arch, Debian and Fedora based distributions.

https://github.com/jtsiros/amd-isp4-camera

I did have an issue building the patch series on Fedora 43. However, I was able to resolve the issue via a simple tweak to the version number. A new forked repository can be found below.

https://github.com/mswbull/amd-isp4-camera

To install on Fedora, simply download the repository and navigate to the directory within the Terminal.

From within the directory, run the following commands:

sudo dnf install kernel-devel kernel-headers gcc make b4
make install

The webcam should now be enabled.

You will need to run the ‘make install’ command each time the kernel is updated. Therefore, I recommend keeping the files local.

M5 Max USB-C Regression

Thu, 26 Mar 2026 00:00:00 -0500

Update - Sam McLeod has written an excellent article on his blog covering the HiDPI limitation in detail.

I recently upgraded my Apple MacBook Pro 14-inch to an M5 Max (specification below).

Apple MacBook Pro 14-inch (MAR-2026)
Apple M5 Max (18-core - 6 Super / 12 Performance)
40-core GPU
16-core Neural Engine
128GB Unified Memory (614GB/s Memory Bandwidth)
2TB SSD (7.4GB/s Read)
Wi-Fi 7 and Bluetooth 6
3x Thunderbolt 5, HDMI 2.1, 3.5mm Headphone, SDXC Card Reader
14.2-inch Liquid Retina XDR Nano-texture Display (3024x1964 @ 120Hz)

This is a monster laptop, delivering unrivalled performance in a compact package.

When working at home, I connect my MacBook Pro to a Samsung G95NC Odyssey Neo G9 display, which has a native resolution of 7680x2160 and a maximum refresh rate of 240Hz.

To minimise the number of cables, I have previously used a Anker Prime Thunderbolt 5 Docking Station, which allowed me to connect my MacBook Pro (and other laptops) via one cable (Thunderbolt 5) delivering power and the full 7680x2160 resolution at 120Hz with HDR enabled.

This has always worked perfectly with pervious MacBook Pro laptops (e.g., M4 Max), connected to the same display and docking station via DisplayPort 2.1.

Unfortunately, I have been unable to get the M5 Max to connect via USB-C at the native resolution above 60Hz. At first, I figured this issue was caused by a software bug or a defective cable. However, after some basic troubleshooting, it became clear that this was a limitation of the M5 Max itself.

Thankfully, the issue does not impact the HDMI 2.1 port, which can deliver a 7680x2160 resolution at 120Hz with HDR enabled. Therefore, I have replaced the docking station and returned to using multiple cables.

Overall, this is a minor inconvenience. However, the outcome is still frustrating, as it limits the ability to use the USB-C ports with high performance displays, which I consider a core requirement for a laptop targeting professional users.

On a positive note, I can see no reason why this issue would be caused by a hardware limitation, assuming Apple have not made a mistake with their latest design. Therefore, I hope this issue can be resolved via a firmware update.

Gartner AI Webinar

Tue, 17 Mar 2026 00:00:00 -0500

Earlier today, I presented as part of a Gartner private webinar, covering our highly successful journey with Artificial Intelligence (AI), specifically the development and scale of an Agentic AI platform.

Outlined below are a few highlights from the webinar, which included a live overview of the platform.

User Adoption: The custom platform, ElancoGPT, boasts a 90% adoption rate across the workforce (~11,900 users) with employees independently creating over 4,000 custom AI personas and agents.
From Chatbot to “Agentic” OS: Elanco evolved a simple 2022 chatbot into a multi-model and multimodal platform that integrates directly with core enterprise systems (e.g., SAP, ServiceNow, SharePoint) and autonomously acts on data.
Tiny Team, Huge Impact: The platform was built entirely in-house by a small squad of fewer than six engineers—many of them students or recent graduates—without the use of expensive third-party consultancies.
Staggering ROI: The custom build generates a conservative $6.8 million return on investment against operating costs of just $120,000 per year. For comparison, licensing Microsoft 365 Copilot for the same workforce would cost roughly $4.5 million annually.
Real-World Value: AI is actively transforming daily operations, including automating weeks of manual R&D regulatory work, instantly triaging HR requests, and creating AI-generated customer insights for sales reps.

It was a fun presentation, with approximately 60 companies in attendance. The session was recorded. Therefore, if you are a Gartner customer, you should be able to request access.

Product Vision

Sun, 01 Feb 2026 00:00:00 -0600

The Silicon Valley Product Group (SVPG) describe a product vision as a persuasive, inspiring, and long-term description of the future state.

It serves as a “North Star” to align and motivate product teams, ensuring they remain focused on the outcome of discovering and delivering solutions that solve the highest-value problems.

Partnered with Google, we recently launched our new product vision for Elanco IT, leveraging the power of Generative AI (Veo and Flow) to highlight the transformative power of technology and data.

The Elanco brand is grounded in real imagery, celebrating our customers as authentic, real people. The use of Generative AI in this case targets our internal workforce. Although not perfect, it helps our product teams imagine the future, providing a common understanding to ensure team alignment.

AI Future

Mon, 01 Dec 2025 00:00:00 -0600

The hype around Artificial Intelligence (AI) continues to build. Is it a fundamental shift for humanity or a bubble that will eventually burst?

In my opinion, it is both. There is no question that we are in a period of heightened excitement, with many over-inflated claims and businesses attempting to profit from the market turbulence.

However, there is also genuine innovation, supported by a perfect storm of computing capability, software sophistication, and data availability, backed by massive investment.

I believe that AI will transform the way we work, impacting every industry. Outlined below are my thoughts as to how this transformation could impact enterprise businesses.

AI agents will become increasingly powerful, with an ability to pursue goals, take action and learn autonomously.

These AI agents will be built on large-scale, general-purpose AI models, trained on vast amounts of data that can be applied to a wide range of tasks. This includes multimodal input and output (e.g., text, images, video), alongside the ability to break down complex problems into smaller steps to arrive at a more accurate and reliable conclusion.

Each AI agent will be grounded and provided with specific context, covering the business (e.g., framing, principles, policies, standards) and domain specialisation. In addition, AI agents will be granted access to specific data sources and tooling, with the appropriate permissions and approval to trigger actions.

This combination of knowledge, reasoning and action will allow an AI agent to learn and complete complex tasks, similar to a human worker.

An AI agent will not work in isolation; instead, they will collaborate, with each AI agent providing input based on their specialisation. Other AI agents will provide oversight (management), orchestrating the different inputs to achieve the desired goal.

This model, which is technically viable today, is a paradigm shift for how work gets done and could act as a force multiplier. This is because, for the first time, the human workforce will work alongside AI, partnering to achieve specific business outcomes.

These new human-centric, AI-powered teams are designed for rapid, on-demand scalability. By harnessing parallel processing, they achieve greater speed and agility, while their ability to ingest and contextualise insights from large, disparate data sources leads to significantly improved decision-making and precision.

A team that includes a human workforce and AI agents will inevitably outperform a team that is only comprised of humans.

This new way of working can lead to fear, with the concern that jobs will be replaced or lost. Although I believe this is true in certain circumstances, in many cases, I think these new human-centric, AI-powered teams will simply become more productive, delivering increased value and a higher return on investment.

Therefore, I believe it is in everyone’s best interest to embrace this future, where humans and machines partner to achieve outcomes. This requires individuals to upskill, with businesses looking to evolve their practices and processes.

The businesses that proactively engage with a clear strategy that embeds and prioritises AI, including the prerequisites regarding expertise and data, will succeed.

This will deliver a competitive advantage, differentiated from their peers and protected from disruption.

Linus Torvalds

Sun, 30 Nov 2025 00:00:00 -0600

The global tech community includes some amazing talent, with awe-inspiring contributions that have shaped the world.

However, there are a few individuals that I consider tech royalty, standing above their peers, with a direct impact on my LifeinTECH.

My list would include Linus Torvalds, John Carmack, Steve Wozniak, Demis Hassabis, Gabe Newell, and Mark Cerny.

Due to the nature of these individuals, they rarely seek out the spotlight. Therefore, in those rare situations, when they do speak publicly, it is worth taking note.

This leads me to the recent interview with Linus Torvalds (creator of Linux and Git) from Linus Tech Tips.

In this 50+ minute video, Linus Sebastian playfully interviews Linus Torvalds, covering a wide range of topics.

It is well worth a watch, as it combines good (tech-centric) entertainment with insightful knowledge from one of the most influential software engineers to have ever lived.

As an avid fan of Linux and long-time user of Git, it is fun to hear his perspective on the state of the projects, as well as his hopes for the future.

In addition, Linus shares that he is still a Fedora user at home, which I can relate to directly, as this is always the Linux distribution I return to.

As outlined in previous articles, NixOS is the Linux operating system I want to use, as I like the use of declarative configuration files for reproducibility, alongside the flexible package management and atomic architecture.

However, NixOS is still a relatively immature operating system, which in my experience leads to more hiccups. These are rarely system-breaking, but they disrupt my workflow, which can become frustrating.

Therefore, I always find myself reverting back to Fedora, which has a very stable foundation, with a clean/open philosophy, delivering vanilla GNOME and robust package management via DNF and Flatpak.

Regardless of the distribution, I do believe Linux on the desktop is entering a period of prosperity, helped in no small part by the contributions from Valve, alongside the inconsistent and (at times) bewildering strategy from Microsoft with Windows.

Google Antigravity

Thu, 20 Nov 2025 00:00:00 -0600

Earlier this week, Google publicly released Gemini 3 and a new AI-powered Integrated Development Environment (IDE), known as Antigravity.

The IDE is built as a fork of Visual Studio Code and supports multiple AI models, primarily Gemini 3 Pro, as well as Anthropic Claude Sonnet 4.5 and open-source models from OpenAI.

Similar to Cursor, Google Antigravity follows an “agent-first” paradigm, which goes beyond a traditional AI code assistant, with multiple specialised agents operating simultaneously alongside a human developer, delivering actions and collaborating autonomously.

The video below highlights Google Antigravity in action, including a simple use case building a flight tracker that pulls from a public API and integrates with Google Calendar.

Antigravity is available for Windows, macOS and Linux, which is great to see! It is also surprisingly “feature complete” for a version one release.

In my testing, I took my previous vibe coding project (an educational web application to support spelling) and triggered the same prompt with Google Antigravity.

Google Antigravity immediately produced a task list, which made it very easy to see how the prompt was being interpreted and delivered.

After a few minutes (impressive performance), the implementation was completed, and Google Antigravity created a simple walkthrough guide.

The web application itself was fairly simple, but fully functional.

It included the key requirement to add/upload custom words.

The spelling test itself achieved the goal of displaying ten spellings.

Each spelling is verified as correct or incorrect, providing a final score.

Overall, I have been extremely impressed by Google Antigravity, which delivers a comprehensive experience, incorporating many of the best features from other AI coding assistants and vibe coding platforms, such as Cursor and Lovable.

I have been most impressed with the tightly integrated and intuitive workflow, where a human developer can trigger actions at speed, with multi-threaded AI agents delivering end-to-end actions from planning through to testing, including third-party integrations via APIs, etc. These steps can be reviewed in simple language, enabling a clear “human in the loop” peer review process, promoting visibility, understanding, and quality assurance.

In addition, Google can directly benefit from their significant investment in AI across software and hardware. For example, Google Antigravity offers a massive context window, with Gemini 3 handling over 1 million tokens natively, allowing Google Antigravity to understand monorepos without truncation. They also provide a very public preview, available for free for all users, including unlimited tab completions and command requests.

Finally, Google are ready for enterprise business engagement, with tenant-isolated environments by default, alongside SOC 2, ISO 27001, and FedRAMP compliance.

In my opinion, this combination of features, performance, scale and business-ready compliance could reshape the market for AI coding assistants and vibe coding platforms.

InkyPi

Tue, 18 Nov 2025 00:00:00 -0600

With two kids, my wife and I spend most of our day acting as their personal assistants and taxi service. With PE kit, outdoor learning, school homework, lunches, afterschool clubs, evening sports, it is a constant battle to ensure they have everything they need for the day ahead.

Currently, we use a shared calendar to keep track of everything. However, during the morning rush, I wanted a dashboard in the kitchen that could provide “at a glance” information without needing to use my phone.

There are plenty of commercial products, such as Skylight. However, these are fairly expensive (Skylight Frame 10-inch = £159) and commonly require a subscription (Skylight Plus Subscription = £29pa). Considering the fairly basic nature of the technology, I question the return on the investment.

This led me to InkyPi, which is an open-source E-Ink display powered by a Raspberry Pi. The image below shows my prototype, showing a list view of my familiar calendar. The Ikea frame is not perfect, but it can be easily improved by inserting a better cutout (hiding the edges of the E-Ink display).

The InkyPi project includes a range of features, made available as plugins. For example:

Text (Hardcoded or AI Generated)
Calendar (Google, Outlook, or Apple)
Clock
Countdown
GitHub Contributions
Images (Folder, URL or AI Generated)
Newspaper/Comic
RSS Reader
To Do List
Weather
Year Progress

The plugins can be configured quickly and easily via a web interface, with settings that allow for plugins to be refreshed automatically and displayed based on specific criteria.

As the project is open source, it is also extensible, with the option to customise any plugin or add new ones based on your specific requirements.

The build consists of a Raspberry Pi Zero 2 W (ideally with Headers), a Pimoroni Inky Impression Spectra 7.3-inch display and an SD Card for the operating system (Raspberry Pi OS). The unit can be powered via a standard micro-USB socket and is easily mounted within a standard photo frame (you may need to cut a gap in the back).

The core components cost £17.50 for the Raspberry Pi Zero 2 W, £79.50 for the Inky Impression 7.3-inch display, and approximately £25 for an SD card, resulting in a total of £122. This is cheaper than many of the commercial products, with no subscription required.

In addition, the Raspberry Pi Zero 2 W is very versatile, with its 1GHz ARM Cortex-A53 CPU, 512MB SDRAM, mini HDMI, and wireless connectivity via Wi-Fi b/g/n and Bluetooth 4.2. Considering it runs Linux, it can be used to host other applications for home automation, etc.

The video below highlights the InkyPi in action, alongside the build process, which should not take more than 20 minutes total time.

The InkyPi GitHub project includes everything you need to get up and running, configured as a series of scripts that automate the process.

Spark Configuration

Fri, 14 Nov 2025 00:00:00 -0600

Last month, I started testing the NVIDIA DGX Spark, the self-described “Grace Blackwell AI supercomputer on your desk”.

As highlighted in my previous article, the NVIDIA DGX Spark is a very specific product, targeting those who are interested in AI development, using the NVIDIA AI hardware/software ecosystem.

I would not recommend it for anyone looking for general-purpose computing or a device for open source AI inference, where there are plenty of cheaper alternatives (like systems configured with the AMD-Ryzen-AI-Max-300).

This article will explain my initial configuration for the NVIDIA DGX Spark running as a headless server, providing a CUDA-enabled AI platform that can be accessed from any location.

The configuration includes the initial setup via SSH, alongside the installation of Tailscale for secure remote access and Sunshine for low-latency desktop remote access streaming.

The decisions to use Tailscale and Sunshine are very deliberate, as they are open source and free to use. They also offer high performance with a low support overhead, which is perfect for a home lab.

Initial Configuration

Start by powering on the NVIDIA DGX Spark by plugging it in and pressing the button on the back of the device.

Frustratingly, NVIDIA decided not to include any lights to signify power status; therefore, without a monitor attached, there is no easy way to see if the device is actually on. You can place your hand on the front of the device, where you should feel a small amount of air flow.

Assuming this is the first time using the NVIDIA DGX Spark, it will automatically create a local Wi-Fi hotspot, with the SSID and password printed on the included “Quick Start Guide”.

Connect to the local Wi-Fi hotspot via another system, which will allow you to open the system setup page via a browser using the link http://spark-a3aa.local.

The system setup page includes a step-by-step wizard, covering configuration items such as the hostname, username, password, network, etc.

Once configured, it will also complete a full system update and reboot. The NVIDIA DGX Spark can now be accessed over the local area network either via the browser (using the previously configured hostname) or using the NVIDIA Sync tool.

If running Windows or macOS, I would recommend installing and using the NVIDIA Sync tool, which provides a simple way to connect and launch common applications on the NVIDIA DGX Spark. For example, a Terminal connection via SSH or the NVIDIA AI Workbench.

The configuration for the NVIDIA Sync tool is very simple, requiring the hostname or IP address, username and password.

Remote Access (Tailscale)

Tailscale is a fast, secure mesh VPN service, using the open source WireGuard protocol. I use Tailscale to access other devices on my home network, such as my UGREEN NAS, etc.

Tailscale works great with all client operating systems (Windows, macOS and Linux), while being easy to install, configure and use. It is also free for personal use.

Using the NVIDIA Sync tool, open the Terminal, which will automatically connect to the NVIDIA DGX Spark via SSH.

Run the following commands to install Tailscale, which will be used to support secure remote access. The final command (‘sudo tailscale up’) will prompt you to connect to your Tailscale account, establishing the remote connection with the NVIDIA DGX Spark.

sudo apt update

sudo apt install -y curl gnupg

curl -fsSL https://pkgs.tailscale.com/stable/ubuntu/noble.noarmor.gpg | \
  sudo tee /usr/share/keyrings/tailscale-archive-keyring.gpg > /dev/null
  
curl -fsSL https://pkgs.tailscale.com/stable/ubuntu/noble.tailscale-keyring.list | \
    sudo tee /etc/apt/sources.list.d/tailscale.list
    
sudo apt update

sudo apt install -y tailscale

sudo tailscale up

With Tailscale installed and enabled, you can now access your NVIDIA DGX Spark using the Tailscale IP address, which can be viewed using the command ‘tailscale status’.

Remote Desktop Streaming (Sunshine)

Sunshine is a self-hosted game stream host for Moonlight. Delivering low-latency, cloud gaming server capabilities with support for NVIDIA hardware encoding.

Although you could access the NVIDIA DGX Spark via RDP or VNC, Sunshine and Moonlight offer higher performance, designed to run on ARM64 with an NVIDIA Blackwell (GB10) Graphics Processing Unit (GPU).

To install Sunshine, I recommend using the script maintained by eelbaz.

Using the NVIDIA Sync tool, open the Terminal, which will automatically connect to the NVIDIA DGX Spark via SSH. Run the following commands.

git clone https://github.com/eelbaz/dgx-spark-headless-sunshine.git
cd dgx-spark-headless-sunshine

sudo ./configure_headless_sunshine.sh

sudo reboot

Once complete, browse to “https://xyz.local:47990/”, which will display the Sunshine web interface. Within this interface, you can configure a username and password, as well as pair the NVIDIA DGX Spark using a shared PIN.

To pair the device, open the Moonlight client application and connect to NVIDIA DGX Spark via the hostname or IP address (local or Tailscale, depending on your location).

You will be prompted with a PIN, which you can add to the Sunshine web interface.

The NVIDIA DGX Spark is now accessible via Sunshine and Moonlight, providing full desktop remote access streaming.

With everything configured, you can now start experimenting with the NVIDIA DGX Spark using the documentation provided by NVIDIA.

Vibe Code Review

Mon, 03 Nov 2025 00:00:00 -0600

In my previous article (Continuing the Vibe), I shared my latest experimentation with the vibe coding platform, Lovable.

I created a simple, single-purpose web application to support my son with his spelling education.

To help appraise the quality of the code produced by a vibe coding platform, I took the codebase and completed an end-to-end automated review using Cursor.

A summary of the findings can be found below.

Critical Security Issues

Unsafe random in Test.tsx line 139: const shuffled = [...allWords].sort(() => Math.random() - 0.5);. Not cryptographically secure and can be biased. Use crypto.getRandomValues() or a Fisher–Yates with an RNG. Also review src/components/ui/sidebar.tsx line 536.
Missing env validation in src/integrations/supabase/client.ts: SUPABASE_URL/SUPABASE_PUBLISHABLE_KEY can be undefined. Add startup checks; consider a guard/constants module.
Overly permissive RLS policies. After migrations, public read/write on users, words, test_scores, and word_attempts is risky.
No user input sanitization beyond .trim(). Add validation and allowlists.
Authentication/client-only auth relies on localStorage; logouts don’t invalidate server-side sessions.

Code Quality and Architecture

Duplicated user management across Home.tsx, Dashboard.tsx, and Test.tsx. Extract a shared hook or component.
No error boundary for app-wide crashes. TypeScript strict settings (noImplicitAny, strictNullChecks, etc.) are off; risks unknown-typed code. Missing QueryClient defaults for refetch, stale time, retries.
Missing loading skeletons in WordPerformanceHeatmap; only plain text.
Unused route Index.tsx with placeholder content. Unused App.css.
Race conditions in Test.tsx line 179 via nested async after setTimeout; prefer async/await.

Performance

No memoization: heavy list/grid renders and inline functions; query results not memoized.
Unbounded lists: words, users, scores, and attempts load fully. Add pagination/cursors.
Possible loop: Home.tsx line 127: loadUsers() on dialog open; enforce a cooldown or debounce.
Large state: full word/user arrays kept in memory with no streaming.
UI queries are parallelizable where missing; e.g., Dashboard.tsx lines 108–136.
Supabase order() without indexes; verify existing indexes.

Reliability and User Experience

File upload in WordManager.tsx accepts any text; enforce a schema and size limits.
Orphaned records: deleting a user leaves scores/attempts; inconsistent dashboard totals.
Race conditions in auto-creation: double inserts if users create concurrently.
No conflict/rate-limiting handling (e.g., duplicate usernames; concurrent test submissions).
Missing network retries; toasts lack deduplication.
Inconsistent error handling (some toasts, some logs, some silent).

Positive Observations

Clean migration structure.
Useful indexes and foreign keys.
Decent component organization.
Solid UI patterns.
Working RLS setup (policies need tightening).
Broad test coverage.

Overall, Cursor provided the following scores.

Security Score: 4/10

Overly open RLS, no rate limiting, weak input validation, and no authentication.

Code Quality Score: 6.5/10

Duplication and missing boundaries; solid structure and user interface.

Performance Score: 5/10

Missing memoization and pagination.

Overall Score: 5/10

Needs tighter security, deduplication, pagination, and strict typing.

In conclusion, I believe this is a fair review and score. The security score is a little misleading, as I had purposely not included user authentication, recognising the use case and target audience. Therefore, the score of 4/10 reflects reality, but is an expected outcome (part of the design).

These results validate my position that vibe coding is very well positioned for rapid experimentation and the development of single-purpose web applications for personal use (where there is minimal risk).

However, when considering the use of vibe coding for business or commercial use cases, I would highly recommend a thorough peer review, with a focus on security and performance at scale.

NVIDIA DGX Spark

Tue, 28 Oct 2025 00:00:00 -0500

After a five-month delay, NVIDIA has finally released the DGX Spark, the self-described “Grace Blackwell AI supercomputer on your desk”.

The specification of the NVIDIA DGX Spark is listed below.

CPU: Arm 20 Core (10 Cortex-X925 + 10 Cortex-A725)
GPU: Blackwell Architecture
CUDA Cores: 6,144 (Blackwell Generation)
Tensor Cores: 5th Generation
RT Cores: 4th Generation
System Memory: 128GB LPDDR5X 8533 16 channels (256 bit)
Storage: 4TB M.2 NVM-e SSD
Network: 2x 100GB/s ConnectX-7 NIC, 1x 10GB/s Ethernet, Wi-Fi 7, Bluetooth 5.4
Display Connectors: 1x HDMI 2.1a
Connectivity: 4x USB-C

The standout components include the 128GB unified memory, delivering 273GB/s bandwidth, the GB10 Blackwell GPU, which provides up to 1 petaFLOP of AI performance at FP4 precision, as well as the dual 100Gb/s ConnectX-7 network interface cards, enabling high-performance distributed computing.

In addition, the NVIDIA DGX Spark supports the NVFP4 precision format, a novel 4-bit floating-point representation developed specifically for NVIDIA’s next-generation inference pipeline.

In theory, a single NVIDIA DGX Spark is capable of running a 200 billion parameter AI model locally. With two connected via NVIDIA ConnectX Networking, AI models of up to 405 billion parameters become feasible.

The NVIDIA DGX Spark is £3,699.98, making it an expensive proposition. By comparision, the Minisforum MS-S1 MAX, equipped with AMD Ryzen AI Max+ 395 and 128GB LPDDR5x is £2,079.00. You can also purchase an NVIDIA GeForce RTX 5090 for £1,799.

The AMD Ryzen AI Max+ 395 has the same unified memory architecture, with a memory bandwidth of 256GB/s (just below the NVIDIA DGX Spark of 273GB/s).

As a result, for basic AI inference (using AI models via Ollama and/or LM Studio), any system equipped with the AMD Ryzen AI Max+ 395 will perform similarly to the NVIDIA DGX Spark. These systems, which are x86-64, will also be more versatile, as they can run Windows and Linux software with no constraints.

The NVIDIA GeForce RTX 5090 has a memory bandwidth of 1,792GB/s, which is far beyond the NVIDIA DGX Spark and AMD Ryzen AI Max+ 395.

As a result, the NVIDIA GeForce RTX 5090 will easily outperform the NVIDIA DGX Spark for AI inference, whilst also delivering market-leading gaming performance. However, with only 32GB of VRAM, the size of the AI models will be restricted.

Therefore, depending on your use case, the NVIDIA DGX Spark may not offer much unique value. In fact, it may be less performant and versatile, whilst being more expensive.

With this in mind, the only people who should consider an NVIDIA DGX Spark are those who are interested in AI development, using the NVIDIA AI hardware/software ecosystem.

With this audience in mind, the NVIDIA DGX Spark provides a compelling entry point to prototype, fine-tune, and deploy large AI models on a desktop computer, using tools and techniques that translate directly to enterprise NVIDIA products and services.

Over the coming weeks, I will be testing the NVIDIA DGX Spark, targeting a range of AI development workloads.

Stay tuned for updates.

Continuing the Vibe

Fri, 24 Oct 2025 00:00:00 -0500

In my previous article “Vibe Coding”, I documented my experience using Lovable, which is a popular vibe coding platform targeting individuals with limited technical experience.

Overall, I was reasonably impressed with the capabilities and quality of the output. Although I would not recommend vibe coding for complex production-scale web applications, I do believe it is well-positioned to support rapid prototyping and simple (single-purpose) personal projects.

With this in mind, I recently attended my son’s parents’ evening (he is ten). A key development area for him is spelling, and we were provided a sheet of words that he must practice and learn.

As any parent will know, getting a child to engage and focus in an area that they find difficult can be a challenge. Although the sheet of paper provides the required context (list of words), it is hardly inspiring and offers no process to actually facilitate learning.

Hence, I decided to take 30 minutes to build a small web application that would gamify the process of learning, based on these specific words. This included the following prompt.

I am looking to create an educational web application to help children with their spelling.

The web application must be simple, intuitive and mobile responsive.

The primary function of the web application is to test the user (a child) across ten pre-defined words. Each test should select ten different words at random from the pre-defined list.

Each test should display a pre-defined word and then prompt the user to spell the word from memory, checking to see if the word was correctly spelt or not. At which point, the next word in the test is displayed until the test is finished.

The tests must be gamified, with each correct answer earning a score, which should be displayed on a dashboard, with specific targets for the user to achieve.

The web application must include a section for the pre-defined words to be defined, which can be manually entered and saved or uploaded via a comma-delimited text file. The web application should be fun and engaging, recognising that it is targeting a child.

Within a few minutes, I had a fully working web application, with all the key functionality available for testing.

Using the Lovable development environment, I was able to quickly verify the features of the web application, taking note of any required changes. I added a feature to allow for multiple users, connected the Postgres database using the native Superbase integration and fixed a couple of small bugs (mostly padding issues with the user interface).

At which point, I hit publish and uploaded the list of words provided by the school. My son was then able to test immediately, using his iPad.

The screenshots below highlight the core features of the web application, starting with the home screen, where the user can select to start a test, view their dashboard or manage the words. The user switching capability is accessible via the menu in the top right.

When a test is started, a single word is displayed. Each test consists of ten words, randomly selected from the pre-defined list.

The user is prompted to spell the word and submit their answer, which includes a fun animation to signal if the answer is correct or incorrect.

Once the test is complete, the results are displayed.

The dashboard includes gamification features, with results, achievements and a history of previous tests.

The ability to add specific words is important, as this allows a parent to input the words specific to their child’s needs. The words can be added manually or uploaded in bulk.

I will not claim my son loves completing the tests. However, it is certainly better than simply reading the sheet of paper or being forced to spell them aloud. It provides my son with an option to work at his own pace, whilst also feeling a sense of reward, as he improves his score, etc.

In my opinion, this is a great example of how vibe coding can be used very effectively. If I were to attempt to create this web application manually, it would have taken significantly longer. To be honest, that effort would outweigh the value. However, thanks to vibe coding, I was able to create a fully working web application with minimal effort in less than one hour total time.

The fact that it is a small, single-purpose personal project makes it perfect for vibe coding (minimal risk) and opens my mind to what other use cases I could target to improve through the use of these capabilities.

LibreOffice

Thu, 23 Oct 2025 00:00:00 -0500

LibreOffice is an excellent productivity suite, available for Windows, macOS and Linux. Most notably, it is free and open-source (licensed under a dual LGPLv3 and MPL 2.0 model), meaning it can be used for personal and commercial purposes without any payment or subscription.

By default, LibreOffice uses the OpenDocument Format (ODF) as its file format, which is an international standard developed by the ISO and IEC. However, it also supports the Office Open XML file format, used by Microsoft Office.

As a Linux user, it is also the most popular and (arguably) most comprehensive productivity suite. It includes applications for word processing (Writer), spreadsheets (Calc), presentations (Impress), vector graphics (Draw), database management (Base), and formula editing (Math).

Recognising that Microsoft Office is not available on Linux as a native application, LibreOffice provides an excellent alternative and is often “bundled” with Linux distributions, such as Fedora, etc.

Unfortunately, in a world dominated by Microsoft Office, compatibility is an important requirement. As someone who also uses Microsoft Office across macOS and Windows (mostly for work), having a consistent user experience is desired.

Thankfully, it is possible to configure LibreOffice on Linux to work “like” Microsoft Office, whilst also achieving acceptable (not perfect) compatibility.

To achieve this outcome, I recommend the steps outlined below.

NOTE: The steps will be different for each Linux distribution. The steps below are for Fedora.

First, if LibreOffice came bundled with your Linux distribution, I would actually recommend uninstalling and reinstalling using the Flatpak version. This will ensure you have the latest features and receive regular/timely updates.

sudo dnf group remove libreoffice
sudo dnf remove libreoffice*

flatpak install flathub org.libreoffice.LibreOffice -y

With the Flatpak version installed, the next task is to ensure font compatibility, recognising that Microsoft Office includes a range of proprietary fonts.

sudo dnf install \   https://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm

sudo dnf install \   https://download1.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm

sudo dnf install mscore-fonts-all -y

sudo dnf install curl cabextract xorg-x11-font-utils fontconfig

sudo rpm -i https://downloads.sourceforge.net/project/mscorefonts2/rpms/msttcore-fonts-installer-2.6-1.noarch.rpm

With the fonts installed, open “LibreOffice Writer” and select “User Interface” from the “View” menu. Now select “Tabbed” and “Apply to All”.

This will update the user interface for all LibreOffice applications to closely match the default Microsoft Office Ribbon interface.

Next, open “Options” and select “LibreOffice > Appearance”. Within these settings, select “Icons” and “Colibre (SVG)” or “Colibre (SVG+ dark)” depending on if you are running light or dark mode.

This will update the default icons across the applications to closely match the Microsoft Office defaults.

Switch to “Load/Save > General” and update the “Default File Format and ODF Settings” to “Always save as” the Microsoft Office Open XML file format. For example, “Word 2010-365 Document (*.docx)”. This setting must be configured per application, specifically Writer, Calc and Impress.

Finally, switch to “Load/Save > Microsoft Office” and ensure all features are selected; this will help with compatibility between LibreOffice and Microsoft Office.

With these settings configured, LibreOffice should look, feel and operate similarly to Microsoft Office.

For more information about the benefits of LibreOffice, check out the article Pocket-lint.

Windows

Wed, 22 Oct 2025 00:00:00 -0500

As outlined in the article “First PC - The Beginning”, I have been using Microsoft Windows for the past three decades.

During that time, I believe I have used every major release of Windows targeting consumers, whilst also spending a year administering Windows Server (NT/2000) for a large enterprise business.

I vividly remember upgrading from Windows 3.11 to Windows 95 using floppy disks (thirteen disks plus the boot disk). I also remember getting to see a pre-release version of Windows XP, which fueled my excitement for the first Windows NT-based operating system specifically targeting consumers.

Throughout this time period, I also used Apple OS X and later macOS, alongside various Linux distributions. However, Windows was always a major part of my computing world, delivering value at home and work.

Therefore, it is with some regret that for the first time in my life, I no longer have Windows installed on any of my core systems. The only Windows device I have left is a Microsoft Surface Laptop, which I use (very rarely) for Work.

Between my two primary laptops, I use macOS and Linux (specifically Fedora), with Linux also installed on my desktop PC. As I consider my common software and usage scenarios, I have reached a point where Linux has moved from being a compromise to being viable and even preferred. This includes gaming, recognising that I do not play competitive multiplayer and therefore have no issues with Windows proprietary anti-cheat software.

In my opinion, Microsoft has completely lost sight of Windows to the point where there is no clear vision, direction, or meaningful innovation.

I recognise the Microsoft priority regarding Artificial Intelligence (AI) and even agree with the strategic direction from a business persective. However, I feel they have allowed this to distract them from their “right to operate” responsibilities, which must include the estimated 1.5 billion devices running Windows.

As a result, Windows (specificlly Windows 11) has become a mismatch of features, that are mostly self-serving (Microsoft-centric, not customer-centric). For example, forced system hardware requirements, mandated software features, unwanted bundled applications and relentless advertisements for Microsoft services.

In addition, the user experience is more disjointed than ever before, with years of technical debt impacting performance and legacy user interface elements scattered across the operating system, which has eroded any sense of quality or attention to detail.

To be clear, Windows has never been perfect, and Microsoft has historically made very public mistakes (e.g., Windows ME and Windows 8). However, even Windows 8 had a vision and direction, even if it was misguided. In short, even during these more difficult periods, I have always felt that Windows offered enough unique value to warrant its existence.

As I consider my current usage of computing systems, I no longer see a logical place for Windows, where it offers clear value. Ironically, it has now become the “compromise” in the way I used to consider Linux.

For example, if I want a highly opinionated system that “just works”, I would turn to macOS, whilst also reaping the benefits of Apple Silicon. For everything else, there is Linux, which has reached a point of maturity where it is “compatible enough”, usually delivering higher performance and a better user experience that also respects community (e.g., prioritising openness, privacy, choice, etc.)

I sincerely hope Microsoft are able to improve the state with Windows. However, I see little evidence of any meaningful attempt. Personally, I still believe they should consider a more dramatic pivot, as outlined in my article “Microsoft Linux”.

Fedora and NVIDIA

Sun, 12 Oct 2025 00:00:00 -0500

In my previous article “Fedora 42 Configuration”, I outlined the software and dependencies I install as part of my standard configuration for Fedora Linux.

The focus of that article was a laptop installation (Framework Laptop 13), which includes an AMD processor with integrated AMD graphics.

As it stands today, an all-AMD system continues to be the simplest way to get Linux up and running, thanks to the open source drivers supported by AMD.

Although I favour AMD processors, I still prefer NVIDIA when it comes to dedicated graphics. The reason is simple, NVIDIA offer the best performance, with the most innovative features, especially when it comes to advanced graphics rendering such as Ray Tracing, as well as their parallel computing platform and programming model, known as CUDA.

Therefore, this article will document the process to install NVIDIA drivers, which can be slightly convoluted, especially if Secure Boot is enabled.

Outlined below is the specification of my custom-built PC, which includes an NVIDIA GeForce RTX 5090.

MSI MAG X870 Tomahawk WiFi
AMD Ryzen 9 9950X 4.3GHz Base / 5.7GHz Boost (16C/32T)
Noctua NH-D15 G2 LBC CPU Cooler
64GB Corsair Vengeance EXPO DDR5 PC5-48000C30 6000MHz RAM
2TB Crucial T705 M.2 PCI-e Gen5 NVM-e SSD
2TB Samsung 990 EVO Plus M.2 PCI-e Gen5 NVM-e SSD
NVIDIA GeForce RTX 5090 Founders Edition 32GB GDDR7
Corsair HX1200i 1200W ‘80 Plus Platinum’ ATX PSU
Fractal Torrent Compact

The process described in this article should work for any system running a modern NVIDIA graphics processing unit (GPU), such as the 20/30/40/50 series.

It should also work for Fedora 42+, where Secure Boot is enabled. As a reminder, Secure Boot is a security feature in a system’s firmware that verifies the digital signatures of software components during the startup process.

As a result, to use Secure Boot with Fedora and an NVIDIA GPU, you must generate and import a key. Without this key, the drivers will fail to initialise correctly.

The following command can be used to confirm your graphics card is being detected. It also provides details regarding the make/model.

/sbin/lspci | grep -e VGA

Prior to any installation, ensure Fedora is up to date.

  sudo dnf upgrade --refresh -y

Ensure the RPM Fusion repositories are installed.

sudo dnf install \   https://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm

sudo dnf install \   https://download1.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm

sudo dnf upgrade --refresh -y

If Secure Boot is enabled, you must install the signing modules and generate and import a key. As part of this process, you will be asked to provide a password.

sudo dnf install kmodtool akmods mokutil openssl -y

sudo kmodgenca -a

sudo mokutil --import /etc/pki/akmods/certs/public_key.der

Upon completion, restart the system, which will prompt the MOK enrollment.

Select “Enrol MOK” > “Continue” > “Yes” and enter the password. You can then select “Reboot”.

Install the NVIDIA driver and cuda/nvdec/nvenc support, ensuring the use of the open kernel space driver.

sudo sh -c 'echo "%_with_kmod_nvidia_open 1" > /etc/rpm/macros.nvidia-kmod'

sudo dnf install akmod-nvidia

sudo dnf install xorg-x11-drv-nvidia-cuda

The installation is complete once the following command returns the driver version as a response (For example, 580.95.05).

  modinfo -F version nvidia

This process may take several minutes. If you receive any other response, simply wait and retry the command.

Restart the system to finalise the installation.

To see more information about the drivers and graphics card installed, use the following command.

nvidia-smi

The following two commands can be useful if you encounter issues. Specifically, recompiling the installed kernel modules (akmods) and rebuilding the initial RAM file system image (dracut).

sudo akmods --force

sudo dracut --force

Finally, if you ever need to uninstall the NVIDIA drivers, use the following command.

  dnf remove xorg-x11-drv-nvidia\*

The NVIDIA drivers are now installed and running.

WinBoat

Sat, 11 Oct 2025 00:00:00 -0500

Is this the year of the Linux desktop?

I have been using Linux since 2001 and on many occasions have felt it was close to becoming a viable mainstream operating system for general consumers, specifically the more mature distributions, such as Ubuntu and my personal choice, Fedora.

A great example of the progress is gaming, which, thanks to the great work from Valve and the open source community, is now very feasible on Linux, using the Proton translation layer.

Except for competitive multiplayer games, which require Windows-specific anti-cheat software, it works surprisingly well, even delivering better performance than Windows, thanks to the lower resource impact and ability to pre-cache assets, etc.

In addition, the most common software (e.g., productivity, collaboration) is now available for Linux or has an open source alternative. For example, I maintain my list of common software, with options available across Windows, macOS and Linux.

Unfortunately, even with all of this progress, there are still a few specific applications that individuals rely upon. Most notably, gaming and access to specific Windows applications, such as Microsoft Office and Adobe Photoshop.

Although there are alternatives, such as LibreOffice, they are not perfect and do not have feature parity with their proprietary Windows counterparts. These applications are also not compatible with translation layers, such as WINE.

To overcome this issue, users can either dual-boot Windows (not ideal) or leverage virtualisation. For example, I use KVM. However, it requires effort to install and configure.

Introducing WinBoat, a new Linux application that aims to run Windows apps on Linux with seamless integration.

It uses a combination of KVM and Docker, which sounds scary, but uniquely automates the installation and configuration.

Once installed and configured, it provides a user interface to navigate and open Windows applications, with pre-configured file sharing access to provide access to your Linux filesystem.

As this is still a virtual machine, accessed via RDP, you can also load the full Windows operating system to install new applications.

Although WinBoat is currently in beta (pre-release), I have been testing on Fedora and have been very impressed. There are still some rough edges and missing features. However, the overall value proposition is excellent, providing a simple and reliable way to run Windows applications on Linux, with minimal effort.

I hope this project continues to mature and grow, as I see this as one of the final steps required to unlock mainstream adoption of Linux on the desktop.

Folding Phones

Fri, 10 Oct 2025 00:00:00 -0500

Over the past few years, I have been experimenting with folding smartphones. Specifically, the Samsung Galaxy Z Fold and, more recently, the Google Pixel Pro Fold.

When I first started using a folding smartphone, there were many compromises. The durability, size, weight and cost were all significant barriers to mainstream adoption.

This lack of adoption directly impacted the software ecosystem, which was not well optimised for devices with larger displays, further limiting the value proposition.

Thankfully, things have progressed, with almost every previous complaint now rectified to an acceptable level.

For example, the table below provides an overview of the core features found in a modern folding smartphone, highlighting the status (readiness for mainstream adoption, based on my own opinion).

In my opinion, the Google Pixel Pro Fold (versions 9 and 10) underscore the potential of a folding smartphone.

I have used both devices as my daily driver and can confirm that they combine excellent hardware with a native and well-optimised software experience (supported by “vanilla” Android).

Although not specific to a folding smartphone, the integration of Google Gemini within Android has further enhanced the overall value proposition.

This is especially true when compared with the Apple iPhone, which feels like it has not received a meaningful upgrade in many years (there comes a point where faster processors and better cameras offer diminishing returns for most people).

With that said, two areas continue to fall below my expectations.

Camera Quality: The cameras included with folding smartphones are commonly lower quality than the non-folding equivalent. For example, the Google Pixel 10 Pro Fold vs. the Google Pixel 10 Pro. These differences are increasingly small, but do still exist, impacted by the depth of the device when unfolded.
Cost: Starting at £1,749, the Google Pixel Pro Fold highlights the premium price of a folding smartphone. To be competitive, this price must drop much closer to £1,000 (or ideally below). It would also be great to see non-flagship folding smartphones, providing an easier entry point for mainstream users.

Thankfully, I think both of these areas are about to improve, driven by the growing likelihood that Apple is preparing to release their first folding smartphone.

The recent release of the Apple iPhone Air is (in my opinion) a clear signal of intent, as the research and development investment provides a logical pathway from the current design to a folding smartphone.

A new form factor would also reinvigorate the iPhone brand, which has been stagnant for many years, relying mostly on gimmicks (new buttons and marketing) to entice upgrades.

Recognising the standard Apple iPhone 17 offers superior features at a lower price, I do not believe Apple expect the iPhone Air to be a top seller.

It does, however, provide a way to test a new ultra-thin design at scale, whilst also recouping some of the research and development cost. This testing will also provide valuable insights when determining what features to prioritise with a folding smartphone.

As a reference, the photo below compares the Google Pixel 10 Pro Fold alongside the Apple iPhone Air. It is easy to see how you could adapt the current Apple iPhone Air design to include a folding display.

The release of a folding smartphone from Apple would be a major milestone, as Apple is still the most well-recognised and powerful brand within the smartphone market.

Their presence (for many) would validate the viability of the form factor, whilst also increasing consumer awareness for folding smartphones thanks to their colossal marketing power.

It would also deliver much-needed competition. For example, although Samsung, Google, OnePlus, and Motorola make folding smartphones, they are all fundamentally Android-based, delivering a very similar experience.

With Apple, the consumer would be offered something different, forcing every existing player to up their game, specifically in areas where Apple are known to be strong (e.g., Cameras).

As with all competition, it will also likely unlock more aggressive pricing, with some manufacturers taking additional risks to secure early market share.

As an advocate for folding smartphones, I am extremely excited to see how this evolves. It will also be interesting to see if the sheer potential for Apple to enter the market triggers a proactive response from any of the existing players, specifically Samsung and Google.

For the latest information regarding Android, including the current state of folding phones, I would recommend Android Authority.

Structural Prediction

Mon, 29 Sep 2025 00:00:00 -0500

In my previous article (Rational Drug Design), I shared a session presented at NVIDIA GTC from Isomorphic Labs.

The focus of the session was Rational Drug Design With AlphaFold 3, where the team provided insights as to how they develop general AI models focused on biology.

As part of our strategic partnership with Google, we have been fortunate enough to work with the exceptional team at DeepMind. We actively leverage their specialist AI models (such as AlphaFold) and collaborate on future opportunities.

Although I expect the DeepMind team to continue to lead the field, delivering “bleeding-edge” capabilities, the proprietary nature of these specialist AI models can be restrictive.

Therefore, as part of our strategy to support drug development, we leverage proprietary and open AI models to predict compound structures and optimise synthesis to accelerate the discovery process.

Recently, we partnered with WWT to test a new multi-model Structural Prediction platform, leveraging the NVIDIA Inference Microservices (NIM) architecture, which is a set of optimised inference microservices that accelerate the deployment of AI models on any NVIDIA-accelerated infrastructure.

Synthesising large molecules like Monoclonal Antibodies (mAbs) without prior structural knowledge is a costly, time-consuming gamble. Without the confirmed binding affinity, effort can be wasted on non-viable candidates.

Therefore, predicting compound structures upfront identifies high-potential drug candidates and eliminates non-starters, drastically cutting the synthesis time and expense.

With this goal in mind, we tested the open source Boltz-2 structural biology foundational model, formulated based on complexes not in the training data.

The results were impressive, with improved scale and time to prediction.

Prediction throughput increased 260-fold from 3 complexes (default baseline) to 780 complexes in parallel between two separate tools.

Prediction speed increased 5.5x on average (default baseline) from 11 minutes per complex to 2 minutes per complex.

Finally, we validated that the accuracy of Boltz-2 is comparable to that of AlphaFold2, while being 52 times faster.

We are currently working on a case study with WWT, where we plan to share more details of our testing.

However, the initial findings support our strategy to industrialise our use and scale of computational science, targeting proprietary and open AI models, to accelerate drug discovery.

Vibe Coding

Mon, 22 Sep 2025 00:00:00 -0500

In 2021, I wrote an article regarding the release of “GitHub Cpilot”, an AI pair programming capability powered by OpenAI Codex.

This was my first glimpse of a world where code could be generated based on user intent, either by describing the desired outcome or referencing previous code. It sparked a lively debate across the software engineering community, regarding the value and ethics of such a capability, especially one that could (in theory) replace/reduce the reliance on specialist skills.

Personally, I was optimistic, as AI-generated code felt like a natural progression for software engineering, accelerating the development process, whilst also providing new ways to verify and test code, without human intervention.

This journey has led us to the concept and term “vide coding”, which was first introduced in February 2025 by Andrej Karpathy, co-founder of OpenAI and former AI leader at Tesla. Vibe coding refers to a coding approach that relies on large language models, allowing programmers to generate working code by providing natural language descriptions.

Inevitably, this has led to the emergence of specialist “AI coding” platforms, which I would loosely split into two categories. Those platforms that are designed to support software engineers with a technical background, and those that target software hobbyists who have an idea but limited technical experience.

A few examples are listed below. To be clear, these platforms can be used to complement each other, for example, Lovable + Cursor, with a source code management system such as GitHub, providing the common ground.

Software Engineers (AI Coding Assistant)

Software Hobbists (Vibe Coding)

I have been investigating a few of the “Software Hobbist” platforms, most notably Lovable, V0, and Figma Make. In some respects, these platforms are the most interesting, as they attempt to produce full-stack applications, targeting individuals with little or no software engineering experience. If successful, these platforms have the potential to broaden the scope of what we consider to be software engineers.

At work, we have been testing V0 by Vercel and Figma Make, as they most closely align with our architecture standards and technology stack. However, at home, I have been testing Lovable.

Lovable is considered the most “user-friendly” vibe coding platform for non-technologists. Therefore, I was intrigued to understand how simple Lovable has been able to make the process and how far it could be pushed.

As a test project, I decided to create a simple application to track my gym activity, covering weights, cardio, etc. In many respects, this is a simple Create, Read, Update, and Delete (CRUD) web application. However, it would likely take a single developer several working days to develop manually (assuming no reference code).

I started with the following prompt, outlining six requirements. I did not pay much attention to the structure or “completeness” of the requirements, as I wanted to see how Loveable would interpret ambiguity.

I am looking to create a web application to track my gym workouts. The key requirements are listed below:

The web application must be mobile responsive.

The web application must have an interface to add new exercises, which could be weights, cardio, etc.

The web application must be able to track sets and reps.

The web application must be able to track the weight of the weights used for relevant exercises.

The web application must track the date/time of the workout.

The web application must include a simple timer on the homepage, allowing for certain exercises or workouts to be tracked by time. The user should be able to set the length of the time, with a buzzer (sound) at the end.

The initial result (one-shot prompting) was impressive, as after a few minutes of “thinking”, I was presented with a mobile-responsive homepage that includes a workout timer (as defined in my requirements).

Lovable successfully generated all sub-menus, which provide the ability to create workouts, add exercises, sets, etc. The major functional requirements were incorporated, with a simple user journey that made navigating the application intuitive.

The code produced was also easy to navigate and read. I later took the opportunity to update the code directly within the Lovable editor, which worked as designed, with the changes being reflected/maintained throughout the development process.

With the basic application already working, I then iterated to add some missing features. For example, features to edit/delete workouts and capture the one rep maximum for each relevant exercise. This was all achieved by simply prompting the requirement, which I did incrementally.

When testing the application, I noticed a few minor visual errors (user interface alignment issues) and a couple of functional bugs. To resolve, I simply prompted Lovable to review and resolve the issues. This worked surprisingly well, with only one example that required me to specifically state how I wanted the issue to be resolved (still described in natural language via a prompt).

Regarding backing services, I connected Lovable to GitHub for source code management and Superbase, which is a Postgres development platform that has become popular with Vibe Coding platforms. The setup was simple, with Lovable handling most of the integration steps. Once configured, Lovable could leverage the database as a backing service and would automatically commit updates to GitHub.

Overall, the entire process took me approximately one hour, including the time to set up and configure Lovable, GitHub, Superbase, etc. At the end of the hour, I had a fully working application, which I was able to publish via the Lovable hosting option. I could have also used a third-party host, such as Netlify, pushed via GitHub.

I have been testing with the Lovable “Pro” plan, which was £185.25 for the year, which includes 100 credits per month. A credit tracks usage within the platform, with a variable cost being assigned based on the action complexity.

My gym tracking application consumed approximately 22.5 credits.

Lovable also includes a basic security review capability, which aims to highlight potential risks. I am pleased they include this feature. However, I would recommend leveraging a third-party tool, integrated with GitHub, for a more thorough analysis.

Overall, I have been impressed with Lovable. The user interface is intuitive, and the results are impressive, at least for a simple CRUD web application.

With that said, my experience and ongoing analysis have highlighted a few advantages and disadvantages of vibe coding.

Advantages

Prototyping: Vibe coding is great for rapid experimentation and protypting, producing a working asset within hours (vs. days). It is also possible to iterate in real-time, taking direct input from non-technical users. I could see this being very useful when completing discovery with the target user base.
Personal Projects: If you are looking to create a simple CRUD web applications for personal use, vibe coding offers a compelling proposition. The speed and simplicity lower the barrier to entry for non-technical users, making application development feasible for a wider audience.
Boilerplate Code: Vibe coding can produce respectable boilerplate code, providing a foundation to build from. This could be valuable when starting a new project, helping to accelerate through the initial development phase, allowing the software engineer to focus on the more complex logic. In addition, for software engineers with specific expertise (e.g., backend development), the use of vibe coding could replace the need for additional human resources (e.g., frontend development), allowing software engineers to produce “full-stack” deliverables.
Education: If you have previous software engineering experience, vibe coding could be used as a method to learn a new programming language and/or framework. Being able to rapidly generate working code that can be used as a reference could save hours of research, whilst also allowing the software engineer to get “hands on keyboard” more quickly, which accelerates learning and improves engagement (motivation).

Disadvantages

Security: At this time, I would not trust any vibe coding platform to generate secure/hardended code, specificlly for key funcionaility such as authentication and authorisation. Security-specific functionality should be custom-developed or imported from a trusted source, including peer review before publishing and after all changes.
Backing Services: Middleware and/or backing service configuration, including the Superbase integration, prioritises functionaility over security. This can leave the application and data vulnerable. Therefore, I would recommend that all backing services be reviewed before publishing and after all changes.
Complex Logic: Vibe coding is capable of generating simple web applications, that support basic CRUD operations. However, more complex logic and/or building from a “brownfield” application (existing codebase) will likely generate inefficiencies and/or errors that could become increasingly difficult to troubleshoot.
Efficiency: To embrace vibe coding effectively, you must accept that the code generated may not adhere to your personal style. This is ok, as it is similar to how different software engineers have certain preferences. However, as a general rule, the best code is the simplest. Vibe coding may not always generate the simplest or most efficient code. With personal projects, this may not cause a concern, but if the application is designed to scale, the performance and support overhead of unoptimised code could become a significant burden.
Hidden Risks: With vide coding, there is a tendency to “fire and forget”, assuming the generated code is “right”, as long as it is functional. This approach could quickly lead to hidden risks, which will likely grow over time.

Overall, I am pleasantly surprised by the effectiveness of vibe coding. Considering the immaturity of this space, it is impressive that a non-technical user could go from nothing to a published web application in a few hours with little/no previous experience.

I am confident this is only the beginning. I expect vibe coding to continue to mature rapidly, enabling greater scope and complexity. I also expect the disadvantages to be reduced over time, with more advanced features and automated verification checks to mitigate common issues/risks.

There is also an interesting opportunity to blend vibe coding with advanced AI coding assistants, like Cursor and Claude Code, etc. In this scenario, I could see engineers using low-barrier platforms such as Lovable as part of the initial discovery, importing the output into an IDE, and using an AI coding assistant to deliver production-ready code.

Finally, a common question is whether vibe coding will replace the need for software engineers. In my opinion, the answer is simple: no, it will not.

The value of a software engineer expands beyond the code itself, which is really just an asset produced as part of the process (similar to how an architect may produce a blueprint as an asset).

The real value is the hundreds of decisions that contribute to the final code, designed to solve a specific problem. This requires curiosity, critical thinking, and strong situational awareness (to understand dependencies, etc.)

In addition, experts will still be required to review and understand code, especially in large enterprise systems, many of which may operate critical infrastructure, services, etc. Therefore, code review skills, which require foundational expertise, will likely become increasingly important.

With this in mind, like many revolutions before, I expect the use of AI within software engineering to evolve how individuals and teams work, replacing certain aspects of the process, but not the underlying skillset.

I will be watching this space closely to see how it evolves!

AI Reality Check

Fri, 19 Sep 2025 00:00:00 -0500

Earlier today, CXOTalk released their latest podcast, which included an interview with Paul Daugherty, Chief AI Advisor at TPG and former Group CEO of Technology at Accenture.

The podcast, titled “AI Reality Check: What Works and What’s New”, provides a great overview of how businesses are transforming with Artificial Intelligence (AI), specifically Generative AI and Agentic AI.

The full episode can be viewed below.

Many of the topics discussed aligned with my own views and experiences. Specifically, outlined below are ten takeaways that I consider critical to increase the probability of success when looking to embed AI as part of a business strategy.

You do not have an AI strategy; you have a business strategy, which AI helps enable. AI must be an embedded part of the business strategy.
The impact of Generative AI and Agentic AI is real, likely to impact every industry across the value chain. AI will penetrate every layer of business structure and demand wholesale process reinvention.
Large Language Models are only the beginning; other branches of AI will also mature quickly, driving additional value.
Focus on transformative use cases, prioritising outcomes over output. AI investments must contribute to “page-one” scorecard metrics for the business.
Implement responsible AI practices, moving beyond principles to industrialised processes. This includes comprehensive risk ranking and mitigation strategies for potential adverse outcomes.
Businesses must establish verifiable AI practices that support rapid evaluation of new concepts/models, establishing trust, credibility and loyalty.
Sales enablement is a common target for AI. However, the real opportunity is when you broaden the scope to include cross-functional processes and data sources.
Cloud and data are prerequisites, specifically data governance/stewardship. However, do not wait for perfection to get started.
Enterprise architects are key, helping to avoid the pitfalls of legacy capabilities such as Robotic Process Automation (RPA), building a foundation that can securely scale.
Expertise is critical, covering those who build AI solutions and those who must consume them. This must directly influence talent acquisition and ongoing organisational change management (including targeted awareness and education).

CXOTalk is an excellent podcast; I highly recommend subscribing.

AMD Ryzen AI Max 300

Sat, 30 Aug 2025 00:00:00 -0500

I recently started testing two systems equipped with the new AMD Ryzen AI Max+ 395, which, in my opinion, is one of the most interesting x86-64 processors ever released.

The specification of the AMD Ryzen AI Max+ 395 can be found below.

AMD Ryzen AI Max+ 395 (16 Cores / 32 Threads)
3GHz Base / 5.1GHz Boost
16MB L2 Cache / 64MB L3 Cache
Radeon 8060S Graphics (40 Graphics Cores)
2900MHz
128GB Unified Memory (LPDDR5x-8000 / 256-bit)

The AMD Ryzen AI Max+ 395 is technically an Accelerated Processing Unit (APU), as it combines a Central Processing Unit (CPU) and Graphics Processing Unit (GPU) on one package, with access to unified (shared) memory.

The CPU is based on the Zen 5 (Strix Halo) architecture, and the GPU is built on the RDNA 3.5 architecture, utilising the TSMC 4nm FinFET process. The total APU includes a configurable TDP (cTDP) ranging from 45W to 120W, making it suitable for both laptops and desktops.

This is the first time that I have seen a high-performance CPU and GPU embedded on a single package, with access up to 128GB of unified memory, delivering 256GB/s of memory bandwidth.

This approach resembles the highly successful Apple M series, with the primary difference being that the AMD Ryzen AI Max+ 395 is x86-64, making it natively compatible with existing Windows and Linux operating systems.

The two systems I have been testing are the HP ZBook Ultra G1a and the Framework Desktop. The specification for each system can be found below.

HP ZBook Ultra G1a

AMD Ryzen AI Max+ 395 3GHz Base / 5.1GHz Boost (16C/32T)
128GB LPDDR5x-8000 Unified Memory
AMD Radeon 8060S Graphics (40 Graphics Cores)
2TB M.2 2280 TLC NVM-e SSD
MediaTek Wi-Fi 7 MT7925 (Wi-Fi 7 and Bluetooth 5.4)
2x Thunderbolt 4, 1x USB-C, 1x USB-A, HDMI 2.1, 3.5mm Headphone
14-inch OLED Display (2880x1800 @ 120Hz)

Framework Desktop

AMD Ryzen AI Max+ 395 3GHz Base / 5.1GHz Boost (16C/32T)
128GB LPDDR5x-8000 Unified Memory
AMD Radeon 8060S Graphics (40 Graphics Cores)
4TB Western Digital SN850X NVM-e SSD
2x Thunderbolt 4, 1x USB-C, 1x USB-A, HDMI 2.1, 3.5mm Headphone
AMD RZ717 Wi-Fi 7 (Wi-Fi 7 and Bluetooth 5.4)
Realtek RTL8126 5Gbit Ethernet
1x HDMI 2.1, 2x DisplayPort 2.1, 2x USB-C, 2x USB-A, 1x 3.5mm
2x Expansion Card Slots

The fact that this APU supports a laptop and desktop form factor is impressive. The only difference is the cTDP, with the HP ZBook Ultra G1a drawing a peak of 70W and the Framework Desktop hitting the full 120W.

This difference in power target highlights the need to balance thermals and battery life for the laptop, constraints that do not exist in a desktop, allowing for the APU to operate at maximum performance.

As you can see from my specification, the systems are both configured with 128GB unified memory. This memory is shared between the CPU and GPU. Therefore, AMD has equipped high-performance modules, delivering 256GB/s of memory bandwidth. This is similar to the Apple M4 Pro (273GB/s) and is a lot higher than the DDR5 memory commonly paired with Intel and AMD processors.

With that said, the memory bandwidth falls behind the Apple M4 Max (546GB/s) and dedicated graphics cards, such as the NVIDIA GeForce RTX 5090 (1,792GB/s). With it being shared, there is also some risk of contention when under heavy load.

To optimise the configuration for specific workloads, it is possible to pre-allocate the amount of memory via the BIOS. For example, when running large local AI models, you would likely want to allocate more unified memory to the GPU (e.g., 96GB), something that would be less relevant for daily usage (where a maximum of 32GB for the GPU would be more than enough).

The use of unified memory is certainly a benefit when it comes to performance, whilst also helping with battery life. However, this design decision means the memory is not upgradable, which means any buying decision must consider the total amount required over the duration of the product’s lifecycle (potentially resulting in a higher initial price).

Operating System

Although both systems support Microsoft Windows, I will be installing and using Linux, specifically Fedora 42.

Using Linux with bleeding-edge hardware can be a risk, as it takes time for the relevant drivers and software optimisations to hit the Kernel.

Thankfully, the AMD Ryzen AI Max+ 395 works great out of the box with Fedora 42, covering the 6.15.x and 6.16.x Kernels. In my previous article “Fedora 42 Configuration”, I shared the specific details regarding my software configuration.

The only hardware issue I have identified during my testing was the integrated webcam in the HP ZBook Ultra G1a. HP does offer a version of Ubuntu 24.04 with an OEM Kernel, which includes webcam support. However, I prefer Fedora and feel confident that webcam support will eventually make its way to the release Kernel. In the interim, I simply use an external USB webcam.

Regarding the AMD Ryzen AI Max+ 395 itself, compatibility, performance, and reliability are all good. The battery life on the HP ZBook Ultra G1a is acceptable. It will not rival a Mac, and at the time of writing, it likely falls behind Windows running the same hardware. However, I expect this to improve as future optimisations are added to the release Kernel.

As is common with Linux running on a Laptop, BIOS configuration can make a big difference regarding reliability and power efficiency. For example, I had some initial challenges with suspend, which I initially thought was a Kernel or distribution issue, but turned out to be BIOS-related.

As for the Framework Desktop, I am pleased to report that everything is working as designed, without the need for any workarounds. As this is a desktop, thermals and power consumption are less of a concern, resulting in a great out-of-the-box experience. Great work from the Framework team!

Hardware

Regarding the hardware itself, both systems are very well built.

The HP ZBook Ultra G1a is certainly a premium laptop, with a dense and rigid construction, which rivals the best I have seen from Dell, Lenovo, Asus, etc.

The touchpad and keyboard are both excellent, with my only minor complaint being the lack of a haptic touchpad. I have also been impressed with the excellent I/O, speakers, microphone, and webcam, all of which are above what I have come to expect from non-Apple laptops.

The same can be said for the display, which delivers a high resolution (2880x1800) and high refresh rate (120Hz), with beautiful colours thanks to the OLED panel.

The Framework Desktop is equally impressive, with a very small form factor and some smart design decisions that ensure thermal performance is perfectly maintained even under sustained full load.

The FlexATX 400W Power Supply Unit (PSU) is a great addition, as this is an area that is usually cumbersome when working with small form factor systems. I am also pleased to see the addition of two M.2 SSD slots (one on the front and one on the back of the mainboard), alongside two easy-to-access expansion card slots on the front.

My only minor frustration is the PCIe x4 slot, which I wish could have been a full PCIe x16 slot (even if bandwidth was reduced). Although it would not be possible to add a dedicated GPU within the default Framework case, the potential to put the Mini-ITX mainboard in another case would have made for an interesting upgrade option down the road.

For example, Minisforum has announced the MS-S1 Max, which also uses the AMD Ryzen AI Max+ 395 and includes a full PCIe x16 slot, alongside USB4v2 Type-C, delivering 80Gbps throughput.

Performance

The GeekBench 6 scores (Single-Core / Multi-Core) for the HP ZBook Ultra G1a and Framework Desktop can be found below.

HP ZBook Ultra G1a - Battery - Balanced = 2578 / 15825
HP ZBook Ultra G1a - Battery - Performance = 2703 / 14033
HP ZBook Ultra G1a - Powered - Balanced = 2590 / 15523
HP ZBook Ultra G1a - Powered - Performance = 2702 / 14410
Framework Desktop - Balanced = 2810 / 20769
Framework Desktop - Performance = 2840 / 22769

Overall, the AMD Ryzen AI Max+ 395 is extremely impressive! These numbers rival the best consumer parts I have used, with the Apple M3/4 Max being a notable exception. With that said, achieving Apple M3/M4 Pro performance is welcome progress for x86-64.

In addition, the Radeon 8060S, with 40 Graphics Cores, achieves NVIDIA GeForce RTX 4060 performance, falling slightly behind the new NVIDIA GeForce RTX 5060. This is an excellent result for an integrated GPU, which avoids the cost, size, power, and thermal implications of using a dedicated GPU. I suspect the Radeon 8060S will be more than enough for most users, even those with specific requirements for local AI, rendering, and gaming.

Unsurprisingly, the Framework Desktop outperforms the HP ZBook Ultra G1a, supported by the higher cTDP. With that said, HP ZBook Ultra G1a performs very well, highlighting the diminishing returns beyond the 70W cTDP.

With Fedora 42, running GNOME 48, the difference between “Balanced” and “Performance” modes appears to bias single-core performance when thermally constrained. With this in mind, when using the HP ZBook Ultra G1a, it would appear that “Performance” mode would offer small real-world benefits when targeting single-threaded workloads.

For more information regarding performance on Linux, I recommend reviewing the excellent articles by Michael Larabel at Phoronix.

In my opinion, Michael continues to deliver the best (most comprehensive) Linux performance analysis.

Conclusion

Overall, I am incredibly impressed with the AMD Ryzen AI Max+ 395.

The performance, reliability, and convenience of having a single package that can deliver excellent results across a wide range of workloads, with support for laptops and desktops, is a major step forward for the x86-64 architecture.

At the right price point, I can see this APU becoming very popular, especially with AI enthusiasts and gamers, especially those looking to balance size and thermal efficiency.

I also hope the success of this APU encourages AMD to deliver more of the same, integrated future architecture improvements, such as RDNA 4, etc.

3D Printing Evolved

Thu, 28 Aug 2025 00:00:00 -0500

I am a huge fan of Prusa, not just their products, but the company and their values. For example, their leadership and commitment to open source (as outlined in their mission below).

“Make enjoyable machines for everyone to use. Open-source is our heart, we strongly believe in the community and in sharing how we do things so others can learn and create. Our goal is for our printers to remain moddable, easily repairable, and produce amazing prints even decades after their initial release. Reprint the plastic parts, flash custom firmware, read through our circuit board designs, or completely modify our printers to meet your specific needs.”

At home and work, I have been using Prusa 3D printers since 2021, most notably the Original Prusa MINI+.

Recently, I was in the market for a new model with three key requirements.

Automated Multicolour Printing
Larger Printing Area (above 180x180x180)
Auto Bed Levelling (ABL)
Integrated Camera
Wireless Connectivity
Mobile Application (iOS and Android)
Integrated Enclosure

This led me to explore the popular Bambu Lab product range, specifically the Bambu Lab P1S 3D Printer with Automatic Material System (AMS).

Based on my analysis, I believe the Bambu Lab P1S 3D Printer with AMS is the best fit for my requirements, especially with the recent discounts.

Unfortunately, Prusa and Bambu Lab are very different companies. Bambu Lab has a more controversial position regarding how they design and build their products, with an emphasis on proprietary hardware and software.

For example, the differences between Prusa and Bamby Lab can be compared to the differences between macOS and Linux, each offering positives and negatives.

macOS, like Bambu Lab, is proprietary, built with vertical integration (software and hardware tightly coupled). This results in a highly opinionated product (minimal flexibility), with the benefit of being fast, reliable, as well as easy to configure, maintain and use.

Linux, like Prusa, is open and therefore highly versatile and flexible. This provides maximum power to the user (right to repair/modify/upgrade), but can be more complex to configure, maintain and use.

I am not stating one is right or wrong or better or worse. The approaches are simply different. As with macOS and Linux, I use both regularly, as each offers unique benefits depending on the use situation and/or use case.

Philosophically, I prefer the open nature of Prusa. However, 3D printing is an area where I am less likely to “tinker”, therefore I certainly value the “it just works” benefits that come with vertical integration.

Therefore, I decided to purchase the Bambu Lab P1S 3D Printer with Automatic Material System (AMS).

Following a few days of testing, I can confirm the Bambu Lab P1S 3D Printer is excellent. Everything feels incredibly polished and refined, with high-quality hardware and intuitive software.

The Auto Bed Levelling (ABL), integrated camera and wireless connectivity are everything I had hoped they would be. It makes printing a trivial task; simply select the model from the mobile app and tap print.

At which point, the Bambu Lab P1S 3D Printer does everything else, without any pre-configuration, etc.

The inclusion of the Automatic Material System (AMS) is also a great feature, allowing for effortless multicolour printing.

I still love Prusa and am excited to see how their products evolve. At work, I plan to continue to use Prusa, as the versatility, alongside the ability to easily repair, modify, and upgrade, are valuable features. However, for personal use, I cannot fault the elegance of Bambu Lab.

Where Prusa has always felt like a product for enthusiasts, Bambu Lab feels like a product for the masses.

CD Projekt Red Redemption

Fri, 15 Aug 2025 00:00:00 -0500

In my article “Cyberpunk 2077”, I provided my perspective regarding the controversial and disappointing launch of the highly anticipated game from CD Projekt Red.

I concluded with the following:

“I feel the launch of Cyberpunk 2077 must become a case study (and warning) for all future game releases. The key learning would be to ensure the marketing and hype do not get ahead of the product itself, which ultimately forced a release that disrespected the CD Projekt Red team, disappointed customers, and resulted in considerable brand damage that may take years to recover.”

As a technologist, this release highlighted a major challenge faced by all software releases, not limited to games.

Specifically, the difficult balance between meeting a deadline and shipping a quality product.

This challenge is especially true for premium software, such as “Triple-A” games, as there is almost always corporate pressure to recoup the significant cost of development. This pressure is compounded when other external forces are applied, such as marketing campaigns and co-promotions, which are often time-bound.

There is also the very real issue of shipping a product to meet the market demand, something that John Carmack has discussed when reflecting on the development of Rage.

John highlights that waiting for the product to be “perfect” can also be detrimental, as the world continues to evolve, including changes in user expectations.

This is where I believe the product-first principles, outlined as part of the Product Operating Model from the Silicon Valley Product Group (SVPG) are very relevant.

For example, the Product Team principles, which include:

Empowered with Problems to Solve: Product teams must be given a problem to solve, rather than solutions to build, and, most importantly, are held accountable for the results. They are empowered to make decisions and are not “order-takers”.
Outcomes over Output: Establish outcome-based goals, leveraging Objectives and Key Results (OKRs), which create focus on the priority, promote curiosity and provide relevant context. Outcomes are the results of actions, not features.
Sense of Ownership: Product teams must have the sponsorship and autonomy to come up with the solutions to the problems (Product Discovery), alongside the responsibility to build them (Product Delivery).
Collaboration: Promote respectful discourse, assuming good intentions when challenging ideas or decisions. Encourage a diverse perspective, but when a decision is concluded, ensure total commitment.

Relating this to Cyberpunk 2077, I believe CD Projekt Red lost sight of these core principles, allowing the external pressure and hype to overshadow the product itself.

Thankfully, over the past five years, CD Projekt Red has been on the road to redemption, with a clear focus on the product itself, demonstrating humility and vulnerability, with clear actions that address previous issues.

Most notably, “Update v2.0” and the Phantom Liberty expansion, which was released in September 2023.

The decision to persevere against the odds is admirable, as it would have been very easy to step away from controversy, shifting gears to a new project without the negative “baggage”.

With the recent release of Cyberpunk 2077 on Nintendo Switch 2 and Apple macOS, I consider the CD Projekt Red redemption arc complete, with a product that fulfils the original vision, delivered at scale across multiple platforms, targeting a wide/diverse audience.

Kudos and congratulations to the team. I suspect the journey has been very challenging. However, I hope the learnings will allow them to proceed with a new sense of purpose, with hard learned lessons about the importance of ensuring the customer is at the centre of every product decision.

Agentic AI - Builder

Fri, 01 Aug 2025 00:00:00 -0500

Following my previous article “Agentic AI in Action”, my team recently released a new feature as part of our multi-model, multi-modal Generative AI framework for business.

We call this feature “Builder”, which allows for self-contained applications to be generated from a single prompt or as part of an Agentic AI workflow.

These self-contained applications can leverage previously defined framing or ingested context from structured or unstructured data. They are also built following pre-defined standards, using an approved technology stack, ensuring they are supportable, secure, private, and compliant (avoiding technical debt).

The power of this feature lies in producing a “rich” user experience, featuring interactive elements that can further enhance how teams automate business processes and interact with data.

For example, the screenshot below is an interactive dashboard, produced from business financial data. Elements can be selected to unlock further insights, providing different views, etc.

Previously, a dashboard of this type would have been created in Power BI or Tableau, likely requiring an individual (or team) with the relevant skills to identify, integrate, build, test, and verify before being made available to the target stakeholder group.

By comparison, this dashboard could be generated in real-time, by anyone with the relevant access to the information, requiring no specialist skill set. The prompt can even be iterated upon, rapidly adding or modifying the content, appearance, or features.

Finally, the self-contained application generated by Builder can be securely and privately shared.

Dashboards are just one very simple example as to how Builder could be used to create value. In testing, we have generated full stack applications, which could be deployed to support specific business use cases.

At the time of writing, we are limiting the ability to generate some of these more advanced use cases, providing the required time to establish the appropriate governance (avoiding application sprawl).

However, the potential is extremely exciting, further enhancing the value proposition of Generative AI and Agentic AI capabilities within a business.

My Setup (Q3 2025)

Thu, 31 Jul 2025 00:00:00 -0500

Update - The Apple MacBook Pro was updated to an M5 Max. The Home Lab was updated to include the NVIDIA DGX Spark.

I switch between three systems (one desktop and two laptops), which are primarily used for productivity, collaboration, software development, photo editing, video editing, virtual labs, gaming, and game development.

My daily driver (home and travelling) is an Apple MacBook Pro 14-inch (MAR-2026), which delivers an excellent balance of performance, battery life and portability. It is also well-positioned to handle tasks that require significant memory and/or memory bandwidth (e.g., local AI Models), alongside the ability to run software natively designed for the Mac, such as Xcode and Final Cut Pro.

The full specification can be found below:

Apple MacBook Pro 14-inch (MAR-2026)
Apple M5 Max (18-core - 6 Super / 12 Performance)
40-core GPU
16-core Neural Engine
128GB Unified Memory (614GB/s Memory Bandwidth)
2TB SSD (7.4GB/s Read)
Wi-Fi 7 and Bluetooth 6
3x Thunderbolt 5, HDMI 2.1, 3.5mm Headphone, SDXC Card Reader
14.2-inch Liquid Retina XDR Nano-texture Display (3024x1964 @ 120Hz)

My second system is a custom-built desktop PC, running Linux (Bazzite, based on Fedora Silverblue). This system is used to run software optimised for NVIDIA, specifically games and their parallel computing platform and programming model, known as CUDA.

The full specification can be found below:

MSI MAG X870 Tomahawk WiFi
AMD Ryzen 9 9950X 4.3GHz Base / 5.7GHz Boost (16C/32T)
Noctua NH-D15 G2 LBC CPU Cooler
64GB Corsair Vengeance EXPO DDR5 PC5-48000C30 6000MHz RAM
2TB Crucial T705 M.2 PCI-e Gen5 NVM-e SSD
2TB Samsung 990 EVO Plus M.2 PCI-e Gen5 NVM-e SSD
NVIDIA GeForce RTX 5090 Founders Edition 32GB GDDR7
Corsair HX1200i 1200W ‘80 Plus Platinum’ ATX PSU
Fractal Torrent Compact

The AMD Ryzen 9950X and NVIDIA GeForce RTX 5090 are high-performance components, delivering excellent performance across a range of workloads (e.g., multi-threading, ray-tracing gaming, machine learning, video editing, etc.).

Finally, I use an HP ZBook Ultra G1a running Linux (Fedora Workstation) for Linux-specific workloads (e.g., Linux Containers, Ethical Hacking, etc.)

The full specification can be found below:

HP ZBook Ultra G1a
AMD Ryzen AI Max+ 395 3GHz Base / 5.1GHz Boost (16C/32T)
128GB LPDDR5x-8000 Unified Memory
AMD Radeon 8060S Graphics (40 Graphics Cores)
2TB M.2 2280 TLC NVM-e SSD
MediaTek Wi-Fi 7 MT7925 (Wi-Fi 7 and Bluetooth 5.4)
2x Thunderbolt 4, 1x USB-C, 1x USB-A, HDMI 2.1, 3.5mm Headphone
14-inch OLED Display (2880x1800 @ 120Hz)

At home, all systems connect to a 57-inch Super Ultra-Wide monitor, specifically the Samsung G95NC Odyssey Neo G9 240Hz Dual UHD Monitor.

The full specification can be found below.

Model: Samsung G95NC Odyssey Neo G9
Panel Size: 57-inch
Panel Type: VA (Quantum Mini LED - 2,392 Zones)
Panel Curvature: 1000R
Aspect Ratio: 32:9
Resolution: 7680x2160
Refresh Rate: 240Hz
Variable Refresh Rate: AMD FreeSync Premium Pro
Response Time: 1ms (GTG)
High Dynamic Range: HDR10+ Gaming, VESA DisplayHDR 1000
Brightness: 420cd/㎡ (Typical), 1000cd/m2 (Peak)
Interface: 3x HDMI 2.1 (HDCP 2.2), 1x DisplayPort 2.1 (HDCP 2.2)

My workspace is minimal, thanks to the exceptional Secret Lab Magnus Pro XL sit-to-stand desk. The monitor is mounted on a BenQ Ergo Arm BSH01, which is a heavy-duty monitor arm that can (just about) support the 15.4 kg weight of the Samsung G95NC Odyssey Neo G9.

Regarding peripherals, I primarily use the NuPhy Air75 V2 (Wisteria Switches), Pulsar X2 v3 eS Mini Gaming Mouse and Insta360 Link 2. The keyboard and mouse were selected as they support multiple Bluetooth profiles, making it simple to switch between devices. The mousepad is a Pulsar Superglide Glass Mousepad v2, which delivers very low friction and accurate tracking.

The photo also highlights my microphone setup, which is a Blue Yeti connected to the Blue Yeticaster that includes the Compass Boom Arm and Radius III Custom Shockmount. The microphone is primarily used for video conferencing, screencasting, webinars, and podcasts.

Additional peripherals include the Audioengine A1 Home Music System w/ Bluetooth aptX, Shure AONIC 5 Sound Isolating Earphones, 8BitDo Ultimate 2 Wireless Controller, and two virtual reality headsets, specifically the Meta Quest 3 and Apple Vision Pro.

As highlighted in the photo below, the desktop PC is mounted to the desk using the Secret Lab Premium PC Mount, with the PC case door exposed, providing easy access to the components.

Accompanying the desktop PC and monitor is a Bambu Lab P1S 3D printer with Automatic Material System (AMS).

Finally, I use a Fractal Design Refine chair, which balances comfort and personalised ergonomics.

Home Lab

Alongside my desktops and laptops, I have a UGREEN NAsync DXP2800 Network Attached Storage (NAS), which acts as a small home server. The server manages local and cloud backups, server-side workloads (Virtual Machines and Docker), as well as media streaming.

The full specification can be found below.

UGREEN NAsync DXP2800
Intel N100 3.4GHz (4C/4T)
8GB DDR5 4800 MT/s
1x eMMC 32GB (OS Installation)
2x SATA, 2x M.2 SSD (Maximum 76TB)
1x SD Card Reader
1x USB-C, 1x USB-A (10Gb/s), 2x USB-A (480Mb/s)
1x HDMI
1x 2.5Gb/s Ethernet

In addition to the base specification, I installed two 4TB Western Digital Red Plus Hard Disk Drives (5400RPM), a single internal 1TB Solid State Drive, and a 2TB SanDisk Solid State Drive connected externally via USB-C.

The Hard Disk Drives are configured in RAID 1 (mirroring), with the internal Solid State Drive reserved for applications (e.g., Virtual Machines, Docker, etc.) The external Solid State Drive provides a third level of removable backup.

Finally, I have an NVIDIA DGX Spark, which offers a home-friendly option to prototype, fine-tune, and deploy large AI models, using the proprietary NVIDIA AI platform (e.g., NVIDIA CUDA) and tools (e.g., NVIDIA AI Workbench

Minecraft UGREEN NAS

Wed, 16 Jul 2025 00:00:00 -0500

I recently upgraded my home storage to a UGREEN NAsync DXP2800 Network Attached Storage (NAS).

Although not high-performance, the hardware is competent, with an Intel N100 3.4GHz (4C/4T) processor, 8GB DDR5 memory (expandable to 16GB), an SSD for high-throughput I/O and 2.5Gb/s networking.

One of the benefits of a modern NAS is the ability to run Docker, providing access to a wide variety of software.

A great example, my son and daughter enjoy the game Minecraft, which has the option for network play. However, you need access to a Minecraft server to host a game.

There are plenty of online options for hosting Minecraft servers. However, they commonly require account creation and/or a monthly subscription. Therefore, the ability to create a local server is compelling.

I use the itzg dedicated server, which is available for the Bedrock and Java editions of Minecraft. The links below are for the Bedrock version.

This dedicated server is made available as a Docker image, which, conveniently, can be deployed to my UGREEN NAS using the following Docker Compose configuration.

version: "latest"
services:
  Minecraft:
    image: itzg/minecraft-bedrock-server
    environment:
      EULA: "TRUE"
      GAMEMODE: survival
      DIFFICULTY: normal
      VERSION: "latest"
    volumes:
      - ./data:/data
    network_mode: host
    restart: unless-stopped

To deploy, simply connect to the NAS via UGOS PRO (online console) and select the “Docker” application. Click “Project > Create”, provide a “Name”, “Storage Path” and copy/paste the Docker Compose configuration.

This will download, deploy, and run the Docker container. You can then connect to the server from Minecraft (PC Edition) by selecting “Servers” and entering the IP Address of your UGREEN NAS. Although not required, the default ports from Minecraft are:

Minecraft Bedrock Edition: 19132
Minecraft Java Edition: 25565

As per the Docker Compose configuration, this will launch a “Survival” game with the difficulty set to “Normal”. Game settings can be pre-configured by adding specific options, using server properties. The full list of options can be found on the Minecraft Wiki.

By default, these Minecraft servers will only be available on the Local Area Network (LAN). However, with Dynamic DNS (DDNS) and port-forwarding, the server can be exposed to the Internet for remote (online) play.

As stated at the start of the article, the UGREEN NAsync DXP2800 NAS is not the highest-performing. Therefore, do not expect to run large/complex Minecraft servers, target approximately 10 users, with moderate view and simulation distances.

Finally, you can enable or disable Docker Containers from UGOS PRO without losing game status or configuration. Therefore, I tend to keep them disabled unless required.

UGREEN NAS

Tue, 15 Jul 2025 00:00:00 -0500

I recently upgraded my home storage to a UGREEN NAsync DXP2800 Network Attached Storage (NAS).

I had previously used a 4-bay Synology NAS, but more recently, maintained carbon copies of my data on a local Apple Mac Studio with external storage attached.

Although the Mac Studio had served me well, it was overkill for my requirements and lacked the inherent benefits offered by Redundant Array of Independent Disks (RAID). I had also been intrigued by UGREEN, as they had recently entered the NAS market in Europe and the US, with some compelling products.

The specification of the UGREEN NAsync DXP2800 NAS can be found below.

UGREEN NAsync DXP2800
Intel N100 3.4GHz (4C/4T)
8GB DDR5 4800 MT/s
1x eMMC 32GB (OS Installation)
2x SATA, 2x M.2 SSD (Maximum 76TB)
1x SD Card Reader
1x USB-C, 1x USB-A (10Gb/s), 2x USB-A (480Mb/s)
1x HDMI
1x 2.5Gb/s Ethernet

The NAS supports JBOD, Basic, RAID 0, and RAID 1 configurations. I selected RAID 1 (mirroring) for my two Hard Disk Drives and configured the internal Solid State Drive for applications (e.g., Virtual Machines, Docker, etc.) It is possible to use the internal Solid State Drive as a cache to speed up read/write performance. However, none of my use cases require this feature. The external Solid State Drive provides a third level of removable backup.

The hardware specification delivers excellent performance for the price, especially when compared against popular alternatives such as Synology and QNAP. UGREEN are also very consumer-friendly, allowing third-party operating systems to be installed, such as TrueNas and Unraid, as well as no hardware storage restrictions (e.g., any Hard Disk Drive manufacturer can be used).

My requirements include local storage redundancy (mirroring), versatile network accessibility, and cloud synchronisation. I also desire the ability to run local virtual machines and Docker containers. Thankfully, the UGOS Pro operating system provided by UGREEN delivers all of these features; therefore, I have no requirement (at this time) to switch to an alternative.

The hardware itself is very well built, with a compact, sturdy metal chassis, offering simple (tool-less) access to the drives (HDD and SSD), alongside a convenient hatch on the bottom for memory upgrades (expandable up to 16GB).

The software is also excellent! It was instantly discovered on my network and (so far), very fast and reliable when accessing via the web admin console or file sharing protocols. The initial setup process was painless, with a wizard that ensures the basic controls are applied, including user management (alongside multi-factor authentication support) and storage configuration.

The initial RAID 0 setup for my two 4TB Hard Disk Drives took approximately six hours to complete (automated process), at which point the storage volume was available for use.

The majority of the settings are pre-configured. However, I made some minor tweaks to the security controls (restricting access based on my desired network configuration). The included “App Center” provides direct access to a host of native utilities. I installed Cloud Drives (for OneDrive synchronisation), Docker, Virtual Machine, Downloads, Security (Anti-Virus) and Sync & Backup.

The Cloud Drives utility was the most important, allowing me to synchronise (real-time) my data with cloud services, such as Microsoft OneDrive and Google Drive.

Although UGREEN offer native remote access via UGREENlink, I chose to configure my own option using Tailscale. There is no native utility for Tailscale. However, thankfully, it can be installed and configured using Docker.

An example of my Docker Compose configuration can be found below.

services:
  tailscale:
    container_name: tailscale
    image: tailscale/tailscale:latest
    restart: unless-stopped
    volumes:
      - ./tun:/dev/net/tun
      - ./lib:/var/lib
    environment:
      - TS_AUTH_KEY=tskey-auth-<KEY>
      - TS_STATE_DIR=/var/lib/tailscale
      - TS_ROUTES=192.168.0.0/24
    network_mode: host
    privileged: true

The “TS_AUTH_KEY” can be created from the Tailscale web admin console, under “Settings > Keys > Auth keys”.

Once deployed and enabled, the UGREEN NAS should automatically connect and appear in the Tailscale web admin console. At which point, I would recommend configuring “Disable key expiry” and ensuring your subnet routes are configured as desired, under the “Edit route settings”.

You can now connect to your UGREEN NAS over Tailscale using the Tailscale IP address.

Overall, I have been extremely impressed with the UGREEN NAsync DXP2800 NAS. It is cost competitive, well built, performant and reliable, which considering its relative immaturity within the NAS market, is an impressive outcome.

Agentic AI in Action

Sat, 12 Jul 2025 00:00:00 -0500

With the launch of ChatGPT, my team began work on a multi-model, multi-modal Generative AI framework for business.

Over the years, this framework has evolved to include Agentic AI capabilities, enabling AI to take action, supporting a wide variety of business use cases.

We are very proud of our work, which has been publicly recognised, including endorsements from Google and Microsoft, alongside the Business Transformation & Operational Excellence Summit (BTOES) industry award for enterprise AI.

I have documented this journey over a series of blog posts.

As we close the year, our Generative AI framework has reached another significant milestone (see the dashboard screenshot below).

With over 9,695 unique users and approximately 3,000 daily interactions, the organic adoption continues to grow at an impressive rate. Since launch, we have seen over 86,000 hours (3,584 days) saved, resulting in an estimated ROI of $4.1M.

Alongside the business numbers, we transact approximately 3 billion tokens per month across ten primary AI models (Google and OpenAI).

Additional statistics include.

~1,500 Custom Personas Created
~44,000 Documents Translated
~250,000 Lines of Code Accepted
~5,000 External Research Projects Completed
~75,000 Images Created

Thanks to our ephemeral cloud-native architecture, this scale is achieved at a cost of approximately $6,000 per month ($72,000 per year), which includes all users, AI models, integrations and data. By comparison, if we were to use Microsoft 365 Copilot for the same number of users, which has specific restrictions regarding AI models, integrations and data, the cost would be approximately $3.5M per year.

Since my last article on this topic, we have introduced a range of new capabilities.

Deep Research - An agentic AI feature that can automatically browse websites, think through its findings, and create insightful multi-page reports. The process covers planning, searching, reasoning, and reporting, making it useful for competitive analysis, due diligence, etc.
Specialist AI Agents - Dedicated AI agents with the required grounding, context and access to execute specific tasks. The AI agents can be engaged individually or as part of a group, configured within a workflow, allowing for more complex outcomes.
Voice Assistant - A real-time voice assistant, enabling conversational AI. Useful for employees without easy access to a laptop, such as road warriors like sales representatives.
Screen Sharing - The ability to screen share, allowing the screen and all open software/data to be referenced as context. Useful for IT support, when troubleshooting an issue or when looking to produce specific steps personalised to the individual’s configuration.
Interoperability - Model Context Protocol (MCP) support, enabling generative AI models to access and utilise external tools, data, and services in a consistent and reusable way. New integrations include core systems of record, such as ServiceNow, etc.
User Experience - Refined user interface that is optimised for a wide variety of devices, specifically smartphones, installed as a Progressive Web App (PWA) for Apple iOS and Google Android.

In my opinion, the most exciting new capability is the release of specialist AI agents, which can execute specific tasks on behalf of a user, including complex outcomes.

For example, we are working on the following Agentic AI workflow for our sales representatives. The image below outlines the concept.

Outlined below is the workflow that triggers AI agents, specifically the “Web Searcher”, “Web Researcher”, etc.

We believe this AI assistant will support five key aspects of the sales process, solving specific problems.

Unification of customer insights, automatically connecting multiple disparate systems and data sources.
Identifying relevant customer insights, covering internal and external references.
Creation of customer-specific tactics, based on relevant sales playbooks.
Output available via multiple channels, including an audio podcast for consumption on the road.
A personal assistant, pre-grounded with customer context, available for real-time data analysis and support.

The outcome should deliver speed, cost reduction and precision, allowing our sales teams to be more effective within the field, whilst also ensuring our customers receive a personalised engagement.

Similar to this example, we are confident that Agentic AI can be applied across the value chain, supporting a wide variety of use cases. If true, these AI agents will augment the human workforce, positively impact efficiency and effectiveness, whilst directly contributing to margin expansion, specifically profitability.

Exciting times ahead! I’ll be sure to share more information as we progress our story.

Fedora 42 Configuration

Wed, 09 Jul 2025 00:00:00 -0500

With the release of Fedora 42 earlier this year, I thought it was time to update my Fedora configuration article.

The purpose of this article is to outline the steps I take to install and configure Fedora, including third-party software and dependencies.

A full list of the software I use can be found in the article “Software List 2024”.

This article was tested on a Framework Laptop 13, which delivers excellent compatibility, performance, and reliability. However, it should work with any x86-64 desktop/laptop that is capable of running Fedora.

I outlined my experience with the latest Framework Laptop 13 in the article “Framework Laptop Ryzen AI”. As a reference, a summary of the specification can be found below.

Framework Laptop 13 DIY Edition
AMD Ryzen AI 9 HX 370 (12C/24T)
128GB Crucial DDR5-5600 RAM
4TB Western Digital Black SN850X NVMe (7.3GB/s Read)
13.5-inch LCD Display (2880x1920 @ 120Hz)

As previously stated, Fedora works great with the Framework Laptop 13. The installation is fast and painless, with all native hardware being detected and operational without any special configuration.

Fedora Configuration

With Fedora installed and running, I start installing my common applications, leveraging Flathub. Flathub is a repository for Flatpak applications, which are containerised to simplify installation, management and compatibility.

The following commands ensure Flatpak is installed and enabled, alongside the individual installation commands for my common software.

flatpak remote-add --if-not-exists flathub https://dl.flathub.org/repo/flathub.flatpakrepo

flatpak install flathub com.bitwarden.desktop -y
flatpak install flathub org.onlyoffice.desktopeditors -y
flatpak install flathub md.obsidian.Obsidian -y
flatpak install flathub org.standardnotes.standardnotes -y
flatpak install flathub com.visualstudio.code -y
flatpak install flathub org.gimp.GIMP -y
flatpak install flathub org.kde.krita -y
flatpak install flathub com.github.PintaProject.Pinta -y
flatpak install flathub be.alexandervanhee.gradia -y
flatpak install flathub org.flozz.yoga-image-optimizer -y
flatpak install flathub com.obsproject.Studio -y
flatpak install flathub com.getpostman.Postman -y
flatpak install flathub org.audacityteam.Audacity -y
flatpak install flathub org.videolan.VLC -y
flatpak install flathub fr.handbrake.ghb -y
flatpak install flathub org.openshot.OpenShot -y
flatpak install flathub fr.natron.Natron -y
flatpak install flathub org.cryptomator.Cryptomator -y
flatpak install flathub org.gnome.Boxes -y
flatpak install flathub org.godotengine.Godot -y
flatpak install flathub com.heroicgameslauncher.hgl -y
flatpak install flathub com.discordapp.Discord -y
flatpak install flathub com.bambulab.BambuStudio -y
flatpak install flathub io.missioncenter.MissionCenter -y
flatpak install flathub io.github.ilya_zlobintsev.LACT -y
flatpak install flathub org.gnome.Extensions -y

Unfortunately, not every application is available as a Flatpak. Therefore, traditional installation methods are required.

Although Firefox is a great web browser, there are times when Google Chrome is required. Thankfully, the official version of Google Chrome can be installed using the commands below.

sudo dnf config-manager --enable google-chrome

sudo dnf install google-chrome-stable

Zed is a popular code editor with a focus on high-performance and collaboration with humans and Artificial Intelligence. It can be installed using the command below.

curl -f https://zed.dev/install.sh | sh

Docker is a popular platform used to containerise and run applications, commonly used for software development. Fedora natively includes Podman, which is also an excellent container engine, compatible with Docker images.

Docker can be installed using the following commands.

sudo dnf -y install dnf-plugins-core
sudo dnf-3 config-manager --add-repo https://download.docker.com/linux/fedora/docker-ce.repo

sudo dnf install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

sudo systemctl enable --now docker

Ollama is a tool that enables you to run and manage large language models locally. It can be installed using the command below.

curl -fsSL https://ollama.com/install.sh | sh

QEMU, Libvirt and Virt Manager are tools that facilitate virtual machine management.

The tools can be installed using the following commands.

sudo dnf install -y qemu-kvm libvirt virt-install bridge-utils virt-manager

sudo dnf install -y libvirt-devel virt-top libguestfs-tools guestfs-tools

sudo systemctl start libvirtd

Tailscale is a zero-config, mesh VPN service that simplifies connecting devices and services securely across different networks.

Tailscale can be installed using the following commands. Once complete, I also install the “Tailscale QS” GNOME extension for easy access.

sudo dnf config-manager addrepo --from-repofile=https://pkgs.tailscale.com/stable/fedora/tailscale.repo
sudo dnf install tailscale

sudo systemctl enable --now tailscaled

sudo tailscale up

Ghostty is a fast, feature-rich terminal emulator that uses platform-native UI and GPU acceleration. It can be installed using the commands below.

dnf copr enable scottames/ghostty
dnf install ghostty

InSync is a cross-platform file synchronisation application, covering Google Drive, Microsoft OneDrive and Dropbox. Download the installer for Fedora from “https://www.insynchq.com/downloads/linux” and install using the commands below (replace the “x” with the specific version downloaded).

sudo dnf install ./insync-x.x.x.xxxxx-fc42.x86_64.rpm

To ensure reliable media playback and font compatibility, I installed the following dependencies. These require RPM Fusion, which is a repository of add-on packages for Fedora that are not bundled as part of the core distribution.

The following command enables the “free” and “non-free” repositories.

sudo dnf install \   https://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm

sudo dnf install \   https://download1.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm

With RPM Fusion enabled, the command below installs common multimedia libraries.

sudo dnf install libavcodec-freeworld

sudo dnf config-manager setopt fedora-cisco-openh264.enabled=1

Microsoft fonts, which improve Microsoft Office compatibility, can be installed using the following commands.

sudo dnf install mscore-fonts-all -y

sudo dnf install curl cabextract xorg-x11-font-utils fontconfig

sudo rpm -i https://downloads.sourceforge.net/project/mscorefonts2/rpms/msttcore-fonts-installer-2.6-1.noarch.rpm

Steam, the popular digital distribution platform, is available as a Flatpak. However, in my experience, it is better to install it via the DNF package manager using the following command.

sudo dnf install steam -y

Although LibreOffice comes pre-installed with Fedora, the Flatpak version includes more timely updates (commonly improving performance and compatibility). The following commands replace the pre-installed (DNF) version of LibreOffice with the Flatpak version.

sudo dnf group remove libreoffice
sudo dnf remove libreoffice*

flatpak install flathub org.libreoffice.LibreOffice -y

Finally, I only use two GNOME extensions, the previously mentioned “Tailscale QS” and “Tiling Shell”, which is a simple tiling manager for GNOME, similar to what is found on Pop_OS!

That’s it! Fedora is now up and running, with the required software and dependencies installed.

Product Operating Model Masterclass

Fri, 27 Jun 2025 00:00:00 -0500

In previous articles, I have shared my team’s journey towards the Product Operating Model, as defined by the Silicon Valley Product Group (SVPG).

As a reminder, the Product Operating Model is a conceptual model, not a methodology, process or framework. It is about moving from output to achieving outcomes, following a set of product-first principles. These principles focus on ensuring any product is valuable, viable, usable, and feasible.

The Silicon Valley Product Group (SVPG) has published three books on the Product Operating Model, which act as a great reference. Specifically, INSPIRED, EMPOWERED, and TRANSFORMED.

Over the past couple of weeks, Christian Idiodi has been running a masterclass with my team across the US, UK and India. This includes an intense three-day workshop (UK photo below) where every member of the team was immersed in the product-first principles.

Christian hosts an excellent podcast called “Product Therapy”, which I highly recommend.

Outlined below are a few of my favourite statements that highlight key aspects of the product operating model.

The Product Operating Model is a conceptual model, not a methodology, process or framework. Therefore, it must be adapted to suit the specific business context.
The product-first principles are highly applicable to a wide range of businesses, teams and circumstances. They provide a consistent foundation to establish trust through common beliefs and behaviours.
At its core, the Product Operating Model helps businesses identify and solve their biggest problems through discovery and delivery.
Discovery and delivery are team activities that require input from cross-functional experts, including Product Management, Product Design, and Technologists (Engineers).
Focus is critical, it is not about what you choose to do. It is about what you choose not to do. Product Leaders must be willing to say “no” frequently, even if the ideas are compelling. To avoid demotivation, ideas should not be forgotten and can be reassessed at the appropriate time.
Teams must learn to love the problem. They must become experts in the problem, with deep research to ensure context and credibility.
Teams must be empowered and provided the environment to operate with autonomy. Empowerment is about providing a team with the opportunity to discover “how” a problem should be solved. Autonomy aims to minimise the constraints/dependencies that incur friction.
Teams must own the problem through discovery, delivery and operations, avoiding handoffs.
Product Leaders must provide a product vision (a compelling narrative for the future), alongside the strategic priorities that support a common outcome. They must ensure appropriate context is shared and understood by all teams.
Product Managers must prioritise discovery over delivery. If they are overly involved in delivery, it is a signal that the right team members were not proactively engaged as part of discovery.
Engineers prioritise delivery over discovery but need to be proactively engaged in the discovery process to support ideation and rapid experimentation.
Discovery must include rapid iteration, exploring the problem from multiple perspectives. It should include low-cost, low-tech experimentation, delivered with the least amount of effort to help confirm assumptions and verify hypotheses.
Fail fast, fail frequently and always learn. True failure only occurs when you stop trying to solve the problem.
Teams must gain insights from multiple data sources. Accepting just one data source opens the risk of unconscious bias.
Businesses can be opportunity-led or crisis-led, the goal is to become opportunity-led. Always be willing to reinvent to remain relevant.

As Chief Technology Officer (CTO) and Chief Information Security Officer (CISO), I have a bias towards engineering. When I think about engineers who embody the product-first principles, I think of John Carmack.

John Carmack is a legendary engineer, with a resume that rivals the very best. He is also curious, humble, and relentlessly passionate about solving big problems. He also perfectly demonstrates principles of product discovery, demonstrating mastery of the topic and working through rapid, low-tech prototypes.

These behaviours are evident throughout his career, most notably his time at Oculus, where he would frequently run experiments using any available parts (low-tech, low-effort).

I look to these examples for inspiration, hopefully helping me to become a more effective leader.

Microsoft Security Case Study

Fri, 06 Jun 2025 00:00:00 -0500

As Chief Information Security Officer (CISO), I was recently interviewed by Microsoft regarding our use of Microsoft Defender, Defender Experts for XDR and Security Copilot.

Yesterday, a summary of this interview was published on the Microsoft website as a case study.

The case study highlights how my team use Microsoft solutions and expertise to enhance our security posture.

Specifically, how this strategy simplifies our architecture, offering improved reliability and efficiency, that results in optimised detection and response, whilst also reducing costs.

The case study also references our use of Azure OpenAI, supporting our Generative AI and Agentic AI ambitions, with the potential to further enhance our security posture through advanced correlation and triage.

Thanks to the Microsoft team for their continued partnership and for taking the time to recognise my team and our work.

BenQ Ergo Arm BSH01

Sun, 25 May 2025 00:00:00 -0500

As part of my home setup, I use the incrediable Samsung G95NC Odyssey Neo G9 240Hz Dual UHD Monitor.

With a panel size of 57 inches, combined with a resolution of 7680x2160 at 240Hz, this monitor delivers the ultimate experience for productivity and gaming (assuming you have a graphics card powerful enough to drive it).

Unfortunately, a monitor of this size is also heavy, weighing 15.4 kg. This weight, combined with the 1000R curvature makes the monitor very front-heavy, placing a lot of stress on any monitor arm.

Historically, I have used the Ergotron HX with HD Pivot. This heavy-duty monitor arm has incredible build quality but was released before the Samsung G95NC. Therefore, even with the HD Pivot installed with maximum tension, the monitor would droop forward.

As a result, I recently switched to the BenQ Ergo Arm BSH01. BenQ states on its website, “It Handles Any Monitor, Any Weight”, up to 20kg.

I was a little sceptical, as the BenQ Ergo Arm BSH01 does not look or feel as robust as the Ergotron HX.

Installation was very easy, using the C Clamp on my Secret Lab Magnus Pro XL sit-to-stand desk. I did not use the included reinforcement plate (as my desk is metal), however, it is nice to see this option in the box.

Once installed, I attached the monitor using the included quick-release VESA mount. This was a two-person job, as it was difficult to lift and align the monitor simultaneously.

Once installed, I had to increase the tension of the gas spring arm and tilt structure. Interestingly, the gas spring arm did not require full tension, indicating that the monitor arm could (potentially) hold an even heavier monitor.

However, this was not an issue with the Ergotron HX either. The tilt structure was a different story, which did need maximum tension. Unbeliebaly, it held! This is where the HD Pivot on the Ergotron HX failed.

The tilt structure on the BenQ Ergo Arm BSH01 does not look or feel that impressive, but it does do the job!

I now have a monitor that feels robust, with no unwanted sinking, drooping, or sagging. It also provides comfort that the monitor won’t simply drop unexpectedly, potentially damaging the panel and/or desk.

Thanks to BenQ for supporting the extreme end of the monitor market (very few manufacturers do). It is also reasonably priced for a heavy-duty monitor arm, at £119.

Generative AI Usage

Tue, 20 May 2025 00:00:00 -0500

Over the past few years, my team has been on a journey to develop and scale a multi-model, multi-modal Generative AI framework for business, aiming to accelerate productivity and unlock insights from data. This includes Agentic AI capabilities, where AI “agents” can autonomously take action and learn.

Although the technology is great, it is only valuable when applied to support a specific outcome.

Even with the mainstream media hype, it is still evident that many individuals struggle to identify how Generative AI might help them personally. Therefore, I thought I would share my common use cases, many of which I believe are applicable to a wider audience.

Summarisation, Comparison and Analysis

Reference unstructured documents (e.g. Word, Excel, PowerPoint, PDF) to summarise, compare and/or analyse the content. Support the consumption of information, helping to highlight key messages, themes and trends. Examples include white papers, blogs, contracts, policies, earnings, etc.

Writing

Reference existing sources (e.g., documents, communications, interviews) to draft and/or critically appraise written material, incorporating personal style and inflections. This includes sentiment analysis of draft material to help improve/refine the messaging for specific audiences. Examples include communications, blog posts, policies, etc.

Research and Reasoning

Reference private and/or public data sources to prepare a detailed research document regarding a specific topic. This could include a concept (e.g., Quantum Computing), company, product or service. Context can be provided to tighten the scope of the research, identifying strengths, weaknesses, opportunities, threats, etc.

Structured Data Insights

Connect to structured data sources, providing a natural language interface to query and analyse data, including the dependencies and/or connections between data sets.

Knowledge Base

Reference a specific knowledge repository to create an expert system focused on a specific topic. Valuable when interrogating large knowledge articles, manuals, readme documentation, processes, procedures, etc.

Coding Assistant

A software development assistant, specialising in a specific programming/scripting language. Supporting code generation, suggestions, reviews, testing, debugging and documentation creation.

Ideation and Problem Solving

A mechanism to support brainstorming, quickly generating hundreds of ideas focused on a specific problem or outcome.

Image and Video Generation

The creation of images and/or videos for use within presentations, blogs, etc. Side note, the header image for this article was AI generated.

At this point, I can confidently state the use of Generative AI acts as a “force multiplier”. I am able to deliver more, with less effort. It also reduces pressure by automating low-value, but often time-consuming tasks, whilst also frequently improving the quality of the output as I have more time to apply critical thinking.

It should be noted, as part of these use cases, that Generative AI does not replace my individual expertise, experience or style. Instead, it augments my thinking, combining the best of human and machine intelligence.

Although I leverage our custom-built capabilities, with access to advanced AI models, the majority of these use cases can be achieved with consumer capabilities, offered by OpenAI ChatGPT, Google Gemini, Microsoft Copilot, etc.

Marks and Spencer Cyberattack

Wed, 14 May 2025 00:00:00 -0500

I recently posted an article about the importance of data protection, specifically unstructured data.

In addition, I have previously shared my perspective regarding FIDO2 and Passkeys, deployed as a mechanism to remove the reliance on passwords, whilst enabling phishing-resistant authentication.

Recognising the ever-evolving threat landscape and growth in advanced social engineering tactics, I believe these two areas should be prioritised by all individuals and businesses. They do not guarantee safety but would offer significant protection against the impact of this common threat vector.

As a recent example, approximately three weeks ago, Marks and Spencer (M&S) was hit by a cyberattack, which disrupted operations at their retail stores, forced them to suspend online sales and disclose the loss of personal data covering their 9.4 million active customers.

For those that do not know, Marks and Spencer is a major British multinational retailer (founded in 1884), selling clothing, beauty products, home products and food.

At the time of writing, Marks and Spencer are still unable to accept online orders, costing them an estimated £43m a week in lost sales, alongside a share drop of 12%. This does not take into consideration the brand damage, legal/compliance ramifications and significant unplanned investment to recover and fortify their infrastructure.

The full details of the attack are yet to be disclosed publically. However, it is understood that the origin of the attack was social engineering, with an adversary contacting the Marks and Spencer IT Service Desk and convincing the support agent to reset a password on an account. With this access, the adversary was able to steal and encrypt data, crippling operations and forcing a ransom demand.

Marks and Spencer have published a dedicated page focused on the cybersecurity incident. The BBC also has additional details, including advice for anyone impacted by the data loss.

I empathise with the IT and Information Security teams at Marks and Spencer and I hope they are receiving the appropriate support as they work through the details of this high-profile incident.

Finally, let this be a warning to everyone, especially businesses. We must all continue to proactively engage and invest in cybersecurity, ensuring protections are “right-sized” against the risks, with a focus on continuous improvement, including awareness and education.

Framework Laptop Ryzen AI

Fri, 02 May 2025 00:00:00 -0500

My team has been partnering with Framework to bring sustainable computing to enterprise businesses, starting with laptops.

At the time of writing, we have deployed approximately 100 laptops, across the UK, US and Germany. This was part of a multi-phase pilot and business preview.

We were not only evaluating the hardware and software but also the processes that support servicing, upgrades, asset management, compliance, security, etc.

The goal of the partnership is to fundamentally change the traditional model for hardware lifecycle management within an enterprise business, delivering greater user flexibility and lower costs, whilst reducing the environmental impact.

My testimonial is still present on the Framework Business website and we are proud of the collaboration.

Unfortunately, although the business preview itself was a success, verifying the business case. We did encounter a higher rate of hardware failures when compared against our baseline (Dell and Lenovo). We were also limited by hardware availability, caused by the turbulence across the silicon industry (e.g., Qualcomm, Intel, AMD).

As a result, we decided to temporarily pause the wider deployment to allow time for the hardware to mature and the market to stabilise. A more detailed explanation of our findings can be found in the article “Framework Paused”.

In addition, outlined below are the articles I have written about Framework, covering business and personal experiences.

A few weeks ago, I received the latest iteration of the Framework Laptop 13, with the AMD Ryzen AI 9 HX 370.

The full hardware specification can be found below.

Framework Laptop 13 DIY Edition
AMD Ryzen AI 9 HX 370 (12C/24T)
128GB Crucial DDR5-5600 RAM
4TB Western Digital Black SN850X NVMe (7.3GB/s Read)
13.5-inch LCD Display (2880x1920 @ 120Hz)
2x USB-C, 1x HDMI (3rd Gen) 1x 1TB Framework Storage Expansion Card

As you can see from the specification, this is a high-performance Framework Laptop 13, with the most performant processor (AMD Ryzen AI 9 HX 370) and 128GB of memory from Crucial. The official Framework specification states up to 96GB of memory. However, I can confirm the 128GB (2x64GB) Crucial DDR5-5600 SO-DIMM kit works without issue.

The laptop also includes the latest incremental enhancements, including the Fingerprint Reader Kit (2nd Gen), Webcam Module (2nd Gen), and Keyboard (2nd Gen). To be honest, I did not notice any specific improvement with these enhancements, but it is still nice to see a constant flow of updates that are fully backwards compatible.

The exterior of the laptop remains unchanged (the photo below is my existing Framework Laptop 13), which in some respects is underwhelming, but is obviously an important part of the Framework architecture to support future servicing and upgrades. Thankfully, the laptop still looks great, with a modern, elegant design.

Framework has expanded its customisation options via coloured bezels and expansion cards. I was lucky enough to have a translucent bezel from the original batch, which looks great! I would love to see this go further, with a full translucent case. However, I recognise this might not be practical.

I run Linux Fedora as my primary operating system. I also have Windows 11 installed on a 1TB Framework Storage Expansion Card (1st Gen).

I am pleased to confirm that the installation of Fedora 42 was perfect, with no errors or missing drivers. Framework has always excelled with Linux support, selecting Linux-compatible hardware and working with the most popular Linux distributions to ensure a great software experience. In my opinion, Framework is the best way to use Linux on a laptop.

Regarding performance, the AMD Ryzen AI 9 HX 370 performs very well. The Zen 5 architecture provides a healthy single-core performance boost, without negatively impacting the battery life or thermals. In my testing, the AMD Ryzen AI 9 HX 370 was consistently quieter than any previous Framework Laptop 13 I have used (AMD or Intel).

The Geekbench 6 scores are outlined below.

Single-Core Score 2,700
Multi-Core Score: 13,587

As expected, the AMD Ryzen AI 9 HX 370 outperforms the previous generation AMD Ryzen 7 7840U, which scored 2,382 (Single-Core Score) and 12,243 (Multi-Core Score). Approximately a 13% increase for single-core and 10% for multi-core.

As part of this upgrade, I also switched to the new 13.5-inch LCD Display, which I would describe as good, but not great. The higher resolution (2880x1920) and refresh rate (120Hz) are welcomed upgrades, especially as a Linux user, as everything scales perfectly (I use 150% scaling). However, the panel is not overly bright or colour-accurate, putting it below the competition at this price point.

The battery life is also average. In my experience, it does not compete against the latest Dell, Lenovo or HP laptops. If you are a hybrid worker, someone who has regular access to power, I would classify the battery life as acceptable. However, if you are a road warrior, with limited access to power, I suspect the battery life will quickly become a point of frustration.

In conclusion, the AMD Ryzen AI 9 HX 370 is a solid upgrade, especially when combined with the new high-resolution, high refresh-rate display.

As a Linux user, I continue to believe the Framework Laptop 13 is the “best” overall option on the market, with great hardware and software support.As a Windows user, the story is more complex, as the benefits offered by the Framework Laptop 13 are quickly overshadowed by the limitations (e.g., battery life), especially when compared to cheaper alternatives from Dell, HP and Lenovo.

Unstructured Data Protection

Mon, 28 Apr 2025 00:00:00 -0500

Data protection has never been more important, and the value of data continues to rise as it is used to train more advanced Artificial Intelligence (AI) models.

In simplistic terms, data can be categorised as structured or unstructured.

Structured Data - Organised and formatted in a predefined way (e.g., discrete data types such as numbers, short text, dates, etc.)
Unstructured Data - Lacks a predefined format and is stored in its native form (e.g., documents, images, videos, etc.)

Within a business, structured data is commonly persisted within enterprise data stores or systems of record. These are usually tightly governed, with roles, processes and tools to support stewardship, quality, privacy, compliance, security, etc.

The data within these stores or systems are usually secured at source, meaning tight controls regarding how and where the data is persisted and processed, alongside specific standards for ingestion and consumption.

Unfortunately, unstructured data is far less predictable. For example, within my business, we store approximately 240 million unstructured files and produces more than 1 million new files every month. We leverage Microsoft Office 365, therefore the majority of these files are Word, PowerPoint, Excel and PDF documents.

To compound this issue, we also send approximately 60,000 emails daily, which can include business-sensitive information and proliferate the uncontrolled sharing of files through attachments, etc.

Historically, it was common for these files to be completely uncontrolled, with no integrated protection. As a result, they are at risk of data loss, either through malicious actions or an accident.

Therefore, we recently implemented a new global data protection programme that has been designed to address these risks. It incorporates a new Information Classification framework and a series of technical controls, which are enforced (mandatory).

The high-level framework is outlined below.

The goal of the Information Classification framework is to ensure all unstructured data is appropriately classified, using a label.

The three tiers are.

Public - Data approved for public release.
Internal - Data used within the company and with trusted partners.
Restricted - Data that is sensitive and/or confidential.

The simplicity of a three-tier structure helps to balance security with productivity, making it easy for individuals to understand and apply.

The Information Classification framework is enforced via specific technical controls, facilitated by the Microsoft security ecosystem, specifically Microsoft Purview and Defender.

For example, technical controls prevent “Restricted” data from being attached and/or sent to unauthorised third parties. Instead, file sharing must occur via a controlled link, and in certain scenarios, will be encrypted at REST, requiring the recipient to authenticate.

In addition, certain files can only be accessed if they meet specific authorisation criteria, which can include the posture (security/compliance status) of the endpoint.

By default, all emails are automatically classified (no individual action required). However, the individual can decide to reclassify, if required. All new or previously unclassified files require the individual to classify them before they can be accessed. This is a one-time task, which takes only a couple of seconds to complete.

Through this process, all newly created emails and files will be classified, with existing files being captured as they are viewed/modified or archived.

This approach was designed to enable pragmatic scale and minimise business disruption, avoiding the need for individuals to retrospectively label all files and leveraging automation to simplify the process wherever appropriate.

A key learning from the implementation is the importance of achieving the right balance between security and productivity. If the Information Classification framework is overly complex or the controls create too much friction, the risk of business disruption may outweigh the benefits of improved security.

As a result, we invested several months running broad pilots (covering a wide range of scenarios), with a strong focus on organisational change management, specifically training and education.

The outcome dramatically improves the security posture of the business, reducing the risk associated with malicious or accidental loss of unstructured data. It also provides a robust foundation for compliance reporting and future enhancements, with all data being proactively labelled.

Minisforum BD795i SE

Sat, 26 Apr 2025 00:00:00 -0500

Over the years, I have built countless computers, many for myself, but also a lot for family and friends.

I started experimenting with computers in 1992 when I was eight years old. My first PC was an Intel i486SX with 4MB of EDO Memory. I have very fond memories of tweaking the hardware and software of this system, especially my first major upgrade, the 3DFX Voodoo Banshee graphics cards, which I consider the most important and influential technology purchase of my life.

My son is now nine years old and the world of technology has shifted dramatically, with smartphones, tablets and high-performance video game consoles. Therefore, the need to “build” a desktop PC is not what it once was. However, like most kids, my son loves gaming (he mostly plays Fortnite, Minecraft and FC 25). Therefore, I consider it a “rite of passage” that we build his first PC together.

Therefore, leveraging some unused components and taking advantage of a few timely second-hand bargains, we recently built the following PC.

Minisforum BD795i SE and Cooler Master SickleFlow 120
AMD Ryzen 9 7945HX 2.5 GHz Base / 5.4GHz Boost (16C/32T)
Crucial DDR5 RAM 32GB (2x16GB) 5600MHz SODIMM
Samsung 980 Pro NVMe M.2 SSD 1TB (7GB/s Read / 5GB/s Write)
Asus Prime AMD Radeon RX 9070 XT OC Edition PCIe 5.0 16GB GDDR6
AMD RZ616 Wi-Fi 6E (802.11ax) and Bluetooth 5.2
Corsair SF850 850W 80 Plus Platinum PSU
Cooler Master MasterBox NR200P
Gigabyte M32U 31.5-inch IPS Display (3840x2160 @ 144Hz)

The highlight of the build is the Minisforum BD795i SE, which we purchased this part new from Amazon for £383.99.

The Minisforum BD795i SE is a unique Mini-ITX motherboard with an integrated processor and heatsink, advertised as a “Mobile on Desktop”.

In short, this means the motherboard is coupled with an integrated (non-upgradable) mobile processor, speciifclly the AMD Ryzen 9 7945HX. This may sound like a negative. However, the AMD Ryzen 9 7945HX is a highly capable Zen 4 processor, with 16 cores, 32 threads and a boost clock of 5.4GHz.

In addition, as this is a mobile part, it is very power and thermal efficient. Therefore, when combined with the large integrated heatsink and 120mm fan, it can run at full performance (no throttling), whilst consuming just 75W.

In my testing, the AMD Ryzen 9 7945HX performs extremely well, with a Geekbench 6 CPU score of 2,903 Single Core / 16,335 Multi Core. These are impressive numbers, especially when compared against my AMD Ryzen 9 9950X (3,402 Single Core / 21,587 Muti Core), but costs £579.98 for the processor alone (not including the motherboard, heatsink, etc.)

The image below is my son installing the 120mm fan bracket, which comes with the Minisforum BD795i SE.

The only other unique aspect of the Minisforum BD795i SE is the use of SODIMM memory (mobile memory, not desktop). We ordered 32GB (2x16GB) from Amazon Resale (Used), alongside the Cooler Master MasterBox NR200P case. Purchasing used components can be a little risky. However, the cost savings are difficult to ignore and I am pleased to report both parts were perfect (like new).

The memory itself is 5600MHz but clocked at 5200MHz with the Minisforum BD795i SE. This is below the “sweet spot” for AMD Ryzen 7000 series. However, the real-world impact (outside of specialist workloads and/or Synthetic benchmarks) is very low.

The Cooler Master MasterBox NR200P is a great Mini-ITX case, very versatile with great build quality and an elegant aesthetic. Build Mini-ITX can be a little fiddly, especially with a large graphics card (not ideal for a first build), but the outcome is well suited for a kid’s PC (small footprint, power efficient, low noise).

The image below highlights my son installing the motherboard into the case.

As highlighted in my recent article, I found myself with a spare AMD Radeon RX 9070 XT graphics card.

My son has no idea how lucky he is to have this graphics card, delivering a level of performance far beyond his requirements. However, the cost and effort to sell/replace is likely more than it is worth, meaning he finds himself with a very capable gaming system, able to deliver high resolution and high frame rates in the most demanding games (and certainly enough for Fortnite). I hope this graphics card will serve him well for many years to come.

The graphics card will be connected to a Gigabyte M32U 31.5-inch IPS display, which has a native resolution of 3840x2160 at 144Hz, with support for G-Sync and FreeSync. The inclusion of multiple HDMI 2.1 ports is also very useful, making it compatible with modern video game consoles (e.g., PlayStation 5 Pro, etc.)

The display is several years old, but still an excellent choice for productivity and gaming, thanks to the IPS panel. Due to its age, it can also be found for a lower price on the used market, we purchased it second-hand from eBay.

Finally, to ensure the AMD Radeon RX 9070 XT has ample power (750W recommended), we selected the Corsair SF850 850W 80 Plus Platinum PSU. This is a small form factor PSU that is ATX 3.1 and PCIe 5.1 certified, with a Zero RPM mode for the integrated 92mm fan.

Fitting the PSU was a little tricky, as the modular cables sit flush against the back of the graphics card. This issue was compounded by the 3x8-pin power requirements for the Asus Prime AMD Radeon RX 9070 XT OC Edition. Thankfully, with some careful manipulation, the cables could be routed under the graphics card without negatively impacting airflow.

Overall, the build process went well. I cannot say my son demonstrated the same level of enthusiasm as me, but it was a fun project to complete together.

The choice of an Mini-ITX build did complicate the process as it does not provide much room for manoeuvrability. However, the outcome was worth the effort, avoiding the need for a full-size ATX system.

The only other minor pain point is wireless connectivity, which does not come as standard with Minisforum BD795i SE. Thankfully, the motherboard supports standard Wi-Fi modules, such as the AMD RZ616 Wi-Fi 6E, which can be purchased for £18 from Framework.

In conclusion, I hope this PC build project sparks the same curiosity as it did for me as a kid! If not, he can at least enjoy playing Fortnite at a high frame rate!

Removing Ads

Mon, 14 Apr 2025 00:00:00 -0500

Earlier this month, I noticed that Disqus, the third-party comment hosting service I use for LifeinTECH, had started posting advertisements.

As I use an ad-blocker, I had not noticed this change in their service (and missed or ignored any communication they may have sent).

By default, Disqus ad-free usage now requires a subscription plan, specifically Plus, Pro or Business. However, thankfully, Disqus does offer a free “Plus” subscription plan, targeting small, non-commercial sites.

LifeinTECH is my personal blog, where I share my thoughts and showcase projects. It is non-profit and is not associated with, or sponsored by any third parties. I have also always attempted to ensure it does not include any online advertisements or web trackers.

Therefore, I sent the Disqus team an email at “publisher-success@disqus.com” and within 24 hours, the team migrated my site to their free “Plus” subscription plan.

At this point, all ads can be removed by clicking the “Start Deactivation” button on the “Ads” Settings page.

Thanks to the Disqus team for offering this option and accepting my request!

Punish Inaction

Mon, 14 Apr 2025 00:00:00 -0500

I believe Artificial Intelligence (AI) will transform the world. This could be the specific use of AI, for example, in the field of life science, or the general use of AI, helping every individual to maximise their potential.

As it relates to business, the recent article from Jeremy Utley (Adjunct Professor at Stanford University) positions the following statement.

“Good companies reward success, punish failure, and ignore inaction. Great companies reward success and failure, and punish inaction.”

The article itself (linked below) is approximately a 10-minute read and well worth the time, it builds nicely on Jeremy’s previous article “The Most Important AI Role Has Nothing To Do With Code”.

https://www.jeremyutley.design/blog/punish-inaction

As it stands today, many businesses are “dabbling” in the use of AI, specifically Generative AI and/or Agentic AI. It is often positioned as a recommendation, which means engagement and adoption can be inconsistent and slow.

As stated in the article, the difference between “recommended” and “required” is subtle, but can be transformative. For example, “encouragement acknowledges the old paradigm while suggesting a new one. Requirement establishes the new paradigm as the baseline.”

I completely agree with this insight, especially in the context of AI. Therefore, I am personally pushing my business to reward success and failure, and punish inaction, with specific requirements that force action.

To achieve this outcome, it is critical to have clearly defined, objective measures. Measures help remove ambiguity and enforce accountability, ensuring every individual has a clear expectation.

The article outlines a three-part framework for implementing accountability:

Individual Usage Demonstration: Any employee should be able to show, on the spot, how they are using AI in their regular workflow. They should be able to share their screen and walk everyone through one meaningful use case from which they are routinely deriving value. This does not mean showcasing major breakthroughs, just evidence of integration.
Calendar Evidence: Where on an individual’s calendar is experimentation happening? Make AI experimentation a scheduled activity, not something to fit in “when there’s time”.
Normalising the Question: When engaging with your team, don’t be afraid to ask the seemingly obvious question, “Have you tried AI?” Repeatedly asking is a great way to normalise AI use. It takes repetition to work AI into our workflows and muscle memory. Asking this simple question regularly helps reassure individuals that this is an expectation.

This framework is a great starting point for any business looking to drive engagement and adoption of AI, whilst also being simple enough that it can implemented with very minimal effort.

If you agree that AI will transform the world, then as a business, the highest risk associated with AI is inaction. Imagine two businesses, one that has fully embraced AI and another that has not. In the first business, every individual has a “force multiplier”, combining the best of human and machine intelligence.

In my opinion, the approach of “punishing inaction” would create a clear differentiator, potentially unlocking a competitive advantage.

Rational Drug Design

Thu, 10 Apr 2025 00:00:00 -0500

I have worked in healthcare and life sciences for the past eighteen years. Therefore, I have spent a lot of time supporting the drug development process, covering discovery and development, preclinical research, clinical research and approvals.

As anyone in the industry will know, this process can take between ten to fifteen years, requiring significant investment.

In my previous article “Life Sciences AI”, I highlighted a few companies that are doing interesting things with technology, specifically Artificial Intelligence (AI), to support drug discovery and development. I also outlined the process (high level), providing basic insight into the different phases.

Earlier this month, NVIDIA hosted their annual GPU Technology Conference (GTC). There were many incredible sessions. However, the session “Toward Rational Drug Design With AlphaFold 3 and Beyond” from Isomorphic Labs resonated strongly with me.

As part of the talk, Max Jaderberg (Chief AI Officer) and Sergei Yakneen (Chief Technology Officer) from Isomorphic Labs shared how they develop general models of biology and translate these into real-world impact.

The talk itself is approximately 35 minutes long, with 10 minutes at the end for questions. If you are interested in healthcare and life sciences, computational science, and/or Artificial Intelligence (AI), it is well worth watching.

https://www.nvidia.com/en-us/on-demand/session/gtc25-s72684/

I am excited to see how these concepts can be applied to deliver the next generation of high-impact drugs to support human and animal health.

Google Gemini Case Study

Tue, 01 Apr 2025 00:00:00 -0500

Earlier today, Google published a case study documenting our work developing and scaling a Generative AI framework for business.

The case study highlights the following value proposition:

Built private, secure generative AI framework with Gemini and Cortex Framework in just a few weeks.
Reduced time spent on routine tasks by 70% with homegrown Gemini-enabled tool and Cortex Framework for data grounding.
$2.3M saved with AI-generated efficiencies in first year.
Real-time data with Google Cloud analytics and AI improves insights for continuous compliance.
50K pages translated in six months with Translation Hub for an estimated cost savings of $500K.

The content for the article was produced several months ago, therefore we have continued to see positive adoption and value, achieving 65,623 hours (2,734 days) saved, resulting in an estimated ROI $3,145,104.

If you want to learn more, I have documented our journey over a series of blog posts.

Congratulations to the team for the public recognition!

Hardware Journey

Thu, 27 Mar 2025 00:00:00 -0500

I have previously written about my first PC (an Intel i486SX) and a deeper dive in to my first 3D accelerated graphics card (a Voodoo Banshee).

Over the years, I have built countless computers, many for myself, but also a lot for family and friends. At the time, I would chop and change components without a second thought, but in hindsight, I wish I had retained some of the old components or at minimum, documented the process.

As a result, this article will attempt to document the core hardware that I used over the years (not including anything I built for someone else).

There is no real value in this information, beyond plotting my journey through computing hardware.

I have split the table below into three parts, processors, graphics cards and miscellaneous, with miscellaneous covering any noteworthy, interesting or unique components.

NOTE: The list will not be 100% accurate, this is all being constructed from my memory.

Processors

Graphics Cards

Miscellaneous

The iiyama Vision Master Pro 454 was an exceptional 19-inch CRT monitor, capable of a 1920x1440 maximum resolution. It weighed 24Kg and took up a lot of space, but it was incredible for productivity and gaming. In many respects, when I “upgraded” to my first LCD, it was actually a downgrade regarding performance and image quality.

The Coolermaster ATC 201 as I still consider it to be one of the most elegant case designs ever created, years ahead of its time.

The Creative Sounds Blaster X-Fi is a dedicated PCI sound card, which is now rarely part of a PC build. I believe this was the last “premium” sound card I ever purchased.

The Hercules Prophetview 920 Pro DVI was (I believe) my first LCD display. As previously noted, it was in some respects a downgrade, but the I was sold on the thin and lightweight construction, alongside the striking design.

The Noctua NH-D15 is a legendary dual tower cooler, which I have used for years. When it comes to air cooling, I continue to believe this is the best all-round cooler, supporting Intel and AMD.

Finally, the Logitech MX Master mouse, covering the original, 2S, 3, and 3S. Although not the highest performing hardware specification, it is an incredible productivity mouse, with great ergonomics and versatile features for Windows, macOS and Linux.

My plan is to update this table over time, including new components, alongside anything additional I recall.

Practical Product Operating Model

Wed, 26 Mar 2025 00:00:00 -0500

The Product Operating Model, by the Silicon Valley Product Group, is a conceptual model, not a methodology, process or framework. It is about moving from output to achieving outcomes, following a set of product-first principles. These principles focus on ensuring any product is valuable, viable, usable, and feasible.

I have previously written about the Product Operating Model across five articles, including some details regarding my business transition.

In this article, I will share how my business has embedded the Product Operating Model.

All work, across all business functions, including foundational systems and platforms, are delivered and supported by a Product Team.

Product Teams are ephemeral scaling up or down based on the business need. All Product Teams are logically grouped under a specific Product Group.

The diagram is an example of our Product Group/Team topology, with the blue boxes highlighting Product Groups and the white boxes highlighting Product Teams.

The Product Group/Team topology will be different for every business, aligned to the value chain.

Every Product Team has a defined scope and focused problem statement (mission) that is measurable.

The primary goal of a Product Team is to discover and deliver a solution to the problem statement.

By default, a Product Team include a dedicated Product Manager, Product Designer and the associated technical resources. The technical resources are primarily allocated from the Engineering team.

The Product Team will have direct access to the relevant stakeholders, which could include Business Partners, Users, Customers, IT Leaders, etc.

A Product Team operates following the principles of DevSecOps. As a result, any solution discovered and/or delivered is supported by the Product Team.

The flow of work is standardised across all Product Teams, as outlined in the diagram below.

To enable practical scale, Engineering will operate as a unified pool of technical resources, including internal employees and external contractors.

Specific technical expertise and experience will be required to support the depth and breadth of the Product Teams. This includes the following disciplines.

Software Engineering
Data Engineering
Platform Engineering
Site Reliability Engineering
System Administration
Technical Analysts

Regardless of the discipline, all technical resources reporting to the Engineering team are titled “Engineer”, with different levels (e.g., Associate, Senior, Principal) differentiating seniority.

As highlighted in the diagram below, engineers will be allocated to a Product Team based on the needs of the Product Team, as defined by the Product Manager.

Engineers can be durable (persistent), living with the Product Team for the duration of the Product Team lifecycle or dynamically assigned (short-lived) based on a specific time-bound need, providing a “boost” to the team.

Internal employees and external contractors will be treated equally, following the same allocation process, with the same expectations.

External partners are leveraged as part of the Engineering team to support scale and to provide access to the required expertise and experience.

A default external partner (managed service) provides economies of scale and consistency, with boutique partners providing access to specialist expertise and/or experience.

A primary goal of the Engineering team is to discover and deliver feasible solutions, ensuring they meet expectations across the following areas.

Architecture Alignment (Avoiding Technical Debt)
Operational Readiness
Availability
Integrity
Performance
Scale
Efficiency
Security, Quality Assurance, Legal and Privacy
Regulatory Compliance

By default, a Product Team must follow the principles of DevSecOps.

With that said, every system that enters production is classified (Platinum, Gold, Silver, Bronze). Specific systems (e.g., Platinum) may require dedicated support teams and processes to ensure the resolution of issues and incidents are appropriately prioritised.

As previously stated, the Product Operating Model is a conceptual model, not a methodology, process or framework. Therefore, every business looking to embrace the product-first principles will need to carefully consider how they embed the concepts.

Our approach aims to ensure standardisation, focus and discipline, which unlocks speed and quality at scale. It also ensures every team has a clear understanding of the problem they are trying to solve and how that is connected to the overall business outcomes.

This connection between the work of an individual within a Product Team and business outcome drives purpose, which is motivating and promotes clear prioritisation.

NVIDIA GeForce RTX 5090

Mon, 24 Mar 2025 00:00:00 -0500

Earlier this month, I upgraded my custom-built PC. I documented this process across two articles.

The impetus for this change was the NVIDIA GeForce RTX 5090 announcement, which felt like a worthy upgrade to my ageing NVIDIA GeForce RTX 3090 (I had skipped the NVIDIA 40 series).

Unfortunately, NVIDIA’s launch was disappointing, with many issues. Most notably, there was limited supply, an uncontrolled ordering process, and failed MSRP pricing.

Due to these issues, it was almost impossible to purchase an NVIDIA GeForce RTX 5090 at MSRP, resulting in aggressive scalping that pushed the price above £3500 (£1500 above MSRP).

In a perfect world, I would select a different graphics card. Unfortunately, there is no comparable alternative to the NVIDIA GeForce RTX 5090. Therefore, I decided to delay my purchase (selecting a AMD Radeon RX 9070 XT), until the market stabilised and/or the pricing improved.

My team recently attended the NVIDIA GPU Technology Conference (GTC). I did not expect any NVIDIA GeForce 50 series graphics cards to be on sale. However, to my surprise, a colleague was able to secure me an NVIDIA GeForce RTX 5090 Founders Edition for under MSRP (£1688.15).

Although still very expensive for a consumer-grade graphics card, this price is arguably a bargain, £250 below MSRP and upwards of £1750 below current availability pricing.

The full specification of the NVIDIA GeForce RTX 5090 Founders Edition can be found below.

Architecture: Blackwell 2.0
Process: TSMC 4N FinFET (5nm)
Clocks: 2017MHz (Base) / 2407MHz (Boost)
Shading Units: 21,760
Texture Mapping Units: 680
Render Output Units: 176
Streaming Multiprocessors Count: 170
Tensor Cores: 680
Ray Tracing Cores: 170
Memory: 32GB GDDR7 (512bit / 1750MHz)
Interface: PCI Express 5.0

In short, this graphics card is a monster, likely pushing the TSMC 4N FinFET (5nm) process to its limit. This also explains the 575W power requirement, an increase of 125W from the NVIDIA GeForce RTX 4090. As a result, NVIDIA recommends a 1000W PSU or greater, with a native 12V-2x6 connection.

In addition to the increase in power, the NVIDIA GeForce RTX 5090 has 21,760 CUDA cores, approximately 33% more than the NVIDIA GeForce RTX 4090 (16,384 CUDA cores).

The NVIDIA GeForce RTX 5090 also transitions to fifth-generation Tensor and fourth-generation RT cores.

These enhancements are combined with 32GB of GDDR7 VRAM, with a 512-bit memory interface, delivering an impressive memory bandwidth of 1,792GB/s. That is a significant increase over the NVIDIA GeForce 4090 (1,008GB/s) and the ultra-fast unified memory found on the Apple M4 Max SoC (546GB/s).

Considering the power requirements, you would assume that the NVIDIA GeForce RTX 5090 would require a large, elaborate cooling solution. Interestingly, the NVIDIA engineers have demonstrated their expertise by delivering a relatively compact, two-slot design, with a split PCB. This design appears to cool the graphics card very effectively, without any drastic compromises regarding the design, size or noise.

In my opinion, the NVIDIA GeForce RTX 5090 is the “best” built graphics card I have ever owned, with a premium design, high-quality materials and a reassuring construction.

Unsprupsingly, this specification delivers exceptional gaming performance, in line with what you would expect from a generational update that leverages the same 5nm process. With that said, it would appear most of this performance benefit is delivered via the additional 125W of power, further indicating the limitations of the TSMC 4N FinFET (5nm).

I suspect NVIDIA anticipated this limitation, prompting their heavy investment and marketing into software improvements, specifically Deep Learning Super Sampling (DLSS) 4, including Multi Frame Generation (MFG).

DLSS 4 leverages a transformer-based AI model for Super Resolution, which is also backwards compatible with the NVIDIA GeForce RTX 40/30/20 series. I consider DLSS (or equivalent) a “must-have” feature, capable of delivering incredible results.

MFG, which can generate up to three additional frames per rendered frame, is more controversial. It is also exclusive to the NVIDIA GeForce 50 series. As advertised, the technology delivers impressive FPS increases. However, it does not improve latency and can impact image quality. Therefore, I would only recommend MFG in slower-paced games (not Twitch shooters) where the base FPS is already consistently above 60.

With my Samsung G95NC Odyssey Neo G9 display at native resolution (7680x2160), in games such as Cyberpunk 2077 and Indiana Jones, I am able to max all settings, including Ray Tracing and Path Tracing (where supported), whilst still achieving an FPS of above 100. With MFG enabled, this number can easily surpass 250.

Outside of gaming performance, my interest also includes the potential for Artificial Intelligence (AI) acceleration, specifically Generative AI. This is where the CUDA core count and memory capacity/bandwidth provide significant value.

The NVIDIA GeForce 5090 supports v12.0 Compute Capabilities, with the 32GB VRAM offering enough capacity to run 12B parameter AI models natively and up to 50B parameter AI models with low quantization.

As a reminder, quantization is a technique that reduces the computational and memory requirements of AI models by representing their weights and activations with lower-precision data types, like 8-bit integers instead of 32-bit floating points, enabling faster and more efficient deployments. Lower bit quantizations represent more aggressive compression, leading to smaller model sizes but potentially higher quality loss.

Finally, the NVIDIA Encoder (NVENC) and Decoder (NVDEC) capabilities have been upgraded (ninth-generation and sixth-generation), with one additional hardware encoder for each, compared to the NVIDIA GeForce 4090. This is great news for creators and anyone working with video, allowing for more simultaneous workloads with minimal performance loss.

Outlined below are some high-level test results compared to my NVIDIA GeForce RTX 3090 and AMD Radeon RX 9070 XT.

NOTE: My CPU performance has degraded over time. Therefore, the CPU-centric tests are showing a minor reduction in performance.

Overall, when considering graphics-centric tests, the NVIDIA GeForce RTX 5090 outperforms my GeForce RTX 3090 by approximately 99% and the AMD Radeon RX 9070 XT by 40%.

The near 100% increase over my NVIDIA GeForce RTX 3090 is certainly welcomed, highlighting the benefit of skipping a generation. These tests are not comprehensive and the performance benefit will vary depending on the workload.

With that said, these are impressive results, demonstrating the raw horsepower of the NVIDIA GeForce RTX 5090, which is further enhanced when you enable software features, such as DLSS 4 and MFG.

In conclusion, although extremely expensive and power-hungry, I am impressed with the NVIDIA GeForce RTX 5090. I certainly would not recommend this graphics card to everyone, but if you are seeking the “best” available, the NVIDIA GeForce RTX 5090 is the current undisputed king.

I am excited to broaden my testing, specifically with complex workloads related to AI.

Working Rhythm

Fri, 14 Mar 2025 00:00:00 -0500

I have previously written a few articles focused on my career, covering topics such as my approach to career planning, a day in the life and the importance of purpose.

Although some of these articles are fairly old they are still relevant today. For example, the specifics outlined in my “Day in the Life” artcile have evolved. However, the approach remains consistent and continues to serve me well.

In my combined role as Chief Technology Officer (CTO) and Chief Information Security Officer (CISO), I have 7 direct reports and supervise approximately 115 employees and 176 contractors. These employees and contractors are distributed across three primary locations, specifically the UK, US, and India.

With over 300 individuals reporting up to me (approximately 46% of all IT), it is important to find a “working rhythm” that promotes discipline and delivery excellence, ensuring regular touch-points that target specific outcomes.

As a result, I thought I would share the type and cadence of these regular touch-points.

The diagram below highlights a single month, with an approximate snapshot of the regular meetings.

Each structured meeting has a specific focus and outcome.

Lead Team SCRUM Meetings (30 Minutes - Weekly) - Short, time-bound lead team meetings that ensure alignment for the week ahead, focused on immediate priorities and barriers.
Joint Leadership Meetings (60 Minutes - Twice Monthly) - Engineering and Information Security leadership meeting, focused on business context, strategic priorities, people development, ensuring individuals are motivated and empowered.
Team Meetings (60 Minutes - Engineering Monthly / Infomation Security Twice Monthly) - All team meetings, promoting strong situational awareness, connecting team priorities with business outcomes, and ensuring individuals have a sense of purpose.
1:1 Meetings (60 Minutes - Twice Monthly) - Direct report 1:1 meetings, focused on personal development and career planning.
Enterprise Architecture (60 Minutes - Twice Monthly) - Office of the CTO, focused on the IT ecosystem strategy, principles and positioning.
IT Lead Team (120 Minutes - Twice Monthly) - Office of the CIO, focused on strategy, alignment and people engagement.

The cadence of the meetings is based on the need and specific audience. For example, engineers value “time at the keyboard”. Therefore, I attempt to limit the time they are forced to spend in meetings.

Inaddition to this regular cadence, I also run ad-hoc “skip-level” check-ins, where I connect with individuals and groups outside of my direct reports, helping to gain additional, unfiltered feedback and insights.

Overall, I believe this working rhythm helps to ensure quality, focus and discipline across the team, supported by a regular and consistent pattern of engagement.

AMD Radeon RX 9070 XT

Fri, 14 Mar 2025 00:00:00 -0500

Update - Following a disappointing launch by NVIDIA, I was fortunate enough to be one of the few who procured an NVIDIA GeForce RTX 5090 from the NVIDIA GPU Technology Conference (GTC) for under MSRP (£1688.15). More details can be found in the post “NVIDIA GeForce RTX 5090”.

I recently upgraded my main custom-built PC. I documented this process across two articles.

The impetus for this change was the NVIDIA GeForce RTX 5090, which felt like a worthy upgrade to my ageing NVIDIA GeForce RTX 3090 (I had skipped the NVIDIA 40 series).

Unfortunately, the NVIDIA launch was highly disappointing, with a wide range of issues. Specifically:

Misleading marketing, unfairly comparing the NVIDIA GeForce RTX 5070 against the NVIDIA GeForce RTX 4090.
Very limited initial availability, compounded by limited ongoing supply.
Uncontrolled ordering process, resulting in automated “bot” purchases for scalping purposes.
Failed MSRP pricing, with the GeForce RTX 5090 selling for excess of £3500 (£1500 above MSRP).
Reports of the controversial 12V-2x6 connector and/or cables overheating (similar to the NVIDIA GeForce RTX 4090).
A quality assurance failure that resulted in graphics cards shipping with missing ROPs, impacting performance.
Driver issues, resulting in instability and “black screen” issues.

Frustratingly, many of these issues (marketing, supply, ordering and pricing) could have been easily avoided with better proactive planning.

The hardware and software issues, although disappointing, are not uncommon with new product launches and it is unclear the “true” number of impacted users (Reddit and YouTube commonly overstate the real-world impact). However, with consumer sentiment already low due to the launch issues, this additional bad press compounded the market reaction.

Personally, I can forgive the mistakes. However, what I struggle to forgive is the lack of communication and respect from NVIDIA. Since the “launch” of the NVIDIA GeForce 50 series, NVIDIA has failed to proactively engage the community, with only the missing ROPs being formally acknowledged.

In short, humility and vulnerability can count for a lot. If you make a mistake, own it, and look to resolve it as soon as possible.

It would appear that NVIDIA has become arrogant, and emboldened by its market position, ultimately at the cost of its most loyal customers (the same customers that built the company’s foundation).

In a perfect world, I would demonstrate my dissatisfaction with NVIDIA by voting with my wallet. Unfortunately, there is no comparable alternative to the NVIDIA GeForce RTX 5090. This lack of direct competition is a major part of the problem, something we have seen over the years with other technology companies, such as Intel, etc.

Therefore, against my better judgment, I still intend to procure an NVIDIA GeForce RTX 5090. However, I refuse to pay over MSRP. Therefore, I could be in for a long wait.

With this in mind, I devised a new “short-term” plan, which resulted in me purchasing the newly announced AMD Radeon RX 9070 XT as a replacement for my NVIDIA GeForce RTX 3090.

This may seem like an odd upgrade, as the performance difference will not be groundbreaking. However, at the current market pricing, I can sell my NVIDIA GeForce RTX 3090 for around the price of the AMD Radeon RX 9070 XT, delivering a good price/performance upgrade.

AMD Radeon RX 9070 XT

I purchased the ASUS Prime Radeon RX 9070 XT OC Edition which has a list price of £635.99, a little above the £569.99 MSRP stated by AMD. Unfortunately, without a reference card from AMD, it is unclear if the AMD MSRP will be honoured.

The ASUS Prime Radeon RX 9070 XT OC Edition is a dual BIOS, triple fan, 2.5 slot design, with a phase-change thermal pad. It includes a minor out-of-the-box overclock, delivering a boost clock of 3030MHz.

It fits perfectly in my Fractal Torrent Compact case, with a length of 312mm.

The full specification of the ASUS Prime Radeon RX 9070 XT OC Edition can be found below.

Architexture: RDNA 4.0
Process: TSMC N4P FunFET (5nm)
Clocks: 2480MHz (Game) / 3030MHz (Boost)
Shader Units: 4096
Texture Mapping Units: 256
Render Output Units: 128
Compute Units: 64
Tensor Core: 128
RT Cores: 64
Memory: 16GB GDDR6 (256bit)
Interface: PCI Express 5.0

Considering the price (assuming near MSRP), the AMD Radeon RX 9070 XT delivers an impressive and well-balanced specification. For example, the use of GDDR6 (instead of the newer GDDR7) is likely a good cost-saving measure, without sacrificing much performance for this class of graphics card. However, I do wish they could have included a little more memory (ideally 20GB) to support larger AI workloads.

In addition, unlike NVIDIA, AMD does not mandate the 12V-2x6 connector, avoiding the potential heat issues and negative press. The ASUS Prime Radeon RX 9070 XT OC Edition comes with three 8-pin connectors, with a 750W PSU recommendation. The three connectors are likely overkill but could offer some benefit when overclocking.

The ASUS Prime Radeon RX 9070 XT OC Edition includes one HDMI 2.1b and three DisplayPort 2.1a with HDCP 2.3 support. As a result, it should be future-proof for high-resolution, high-refresh-rate monitors.

The default performance of the AMD Radeon RX 9070 XT is impressive. However, I also configured a minor under-volt/overclock (screenshot below), which worked perfectly in my testing and should be viable for 24x7. This can be done by tweaking just three settings in the AMD Radeon Adrenalin Edition application.

Voltage Offset (mV): -80
VRAM Max Frequency (MHz): 2800
Power Limit (%): 10

Outlined below are some high-level test results compared to my NVIDIA GeForce RTX 3090.

Overall, the AMD Radeon RX 9070 XT outperforms my GeForce RTX 3090 by 23%. Although my tests are far from comprehensive, I do believe they provide a basic insight into the performance benefits of the AMD Radeon RX 9070 XT.

I also tested some non-synthetic scenarios, which also delivered impressive results, with a dramatic increase in ray tracing performance compared with RDNA 3.0, etc. When combined with the improvements delivered by FSR 4.0, the AMD Radeon RX 9070 XT is an incredibly well-rounded graphics card.

In summary, based on my initial testing, if you are in the market for a medium/high-end graphics card, targeting gaming only, I would highly recommend the AMD Radeon RX 9070 XT. It delivers compelling performance across a range of work workloads at a compelling price point.

CTO Craft Con

Tue, 11 Mar 2025 00:00:00 -0500

Over the past two days, I attended CTO Craft Con: London, which is a popular gathering for Chief Technology Officers (CTOs) and aspiring tech leaders, with a focus on practical insights and meaningful connections.

The conference started with Erica Stanley (Director of Engineering at Google), exploring how technology leaders can navigate the intersection of rapid technological advancements and global connectedness.

Erica covered key topics, such as how to lead with a focus on ethics, alongside how the role of the CTO can drive business transformation.

Over the two days, there were a series of keynote presentations (outlined below), alongside networking sessions, conversations (round tables) and showcases.

Day One

Mitigating the Hidden Risks of Emerging Tools - Glyn Roberts, CTO, Vention
Bootstrapping the CTO Role: The Best First 100 Days in a New Leadership Position - Roxana Diaconescu, CTO, SilverRail
The Role of the CTO in Shaping Company Culture - Panel (Carwow, Rest Less, Domestic & General, FundamentalVR)
The Future of R&D Productivity: Leveraging a Strategic Framework and GenAI for Lasting Impact - Will Lytle, COO, Plandek
The Connective Tissue of Tech: Why Glue Work Deserves Your Attention - Katja Obring, Director, Kato Coaching Ltd

Day Two

The CTO Spectrum: Which Archetype Are You? - Pat Kua, CTO Coach and Founder, Tech Lead Academy
People, Process, Technology, Business - Anna Shipman, CTO, Kooth
Keeping up with AI - Jon Topper, Strategic Advisor, Ten10 and Founder of The Scale Factory
AI Ethics and Practical Applications: Moving Beyond the Hype - Panel (Dojo, Bumble, Philipps & Byrne, YLD)
Building a Data-Driven Engineering Culture - Oge Opara-Nadi, VP Engineering, Hey Savi

To review the full agenda, please refer to the CTO Craft Con website.

Although there was a lot of great content and discussion, I want to highlight three topics that were most interesting to me.

Will Lytle highlighted the evolving expectations regarding engineers, moving away from output to outcomes. He outlined four pillars that enable organisations to track and optimise value delivery.

He also shared practical examples of useful measures, including measures to avoid, such as traditional volume-based metrics (e.g., lines of code written, etc.)

Katja Obring focused on “glue work”, which is the hidden and commonly under-recognised work that occurs across teams. This work, although not at the top of any priority list, can be business-critical and therefore, must be managed appropriately.

Katja positioned that “glue work” must be made visible and prioritised alongside other key initiatives and the individuals delivering these essential tasks have the relevant sponsorship to succeed and are openly recognised. For example, ensuring “glue work” is fairly distributed, avoiding extreme delegation to a small group of individuals.

In the video below, Katja presents on this topic.

Pat Kua covered the many “archetypes” of a CTO, including engineering-centric, strategic-centric and operational-centric, etc. He challenged the audience to self-appraise, looking to understand personal ambition and perception, alongside what the business may require.

This is a topic I have personally explored over the years, which I covered in my article “CTO”.

Understanding the archetype of the CTO is important, especially for any aspiring CTO. There is no “good” or “bad” archetype. However, to ensure success, it must align and support the specific business outcomes. For example, if a business is heavily focused on a strategic outcome, selecting an operational-centric CTO could become a point of friction and frustration, reducing the probability of success.

Overall, CTO Craft Con was a great event, with a diverse and inclusive community. If you are a current/future CTO or technology leader, I highly recommend checking it out!

Linear

Wed, 05 Mar 2025 00:00:00 -0600

I have previously written about the Product Operating Model across four articles, including some details regarding my business transition.

Linear is a purpose-built tool for planning and building products. Therefore, it is a great companion for any business embracing the Product Operating Model, specifically the product-first principles.

Linear is similar to Atlassian Jira. However, unlike Jira, which over the years has grown in scope and complexity, Linear is very streamlined and highly opinionated.

Although this may seem restrictive, forcing standardisation unlocks quality, speed, agility and scale. Tools that offer greater flexibility require incredible discipline to manage at scale, usually resulting in chaos that commonly impacts quality and speed.

I would recommend reviewing “The Linear Method”, which outlines their philosophy for building quality products.

Getting started with Linear is fairly simple and the team have produced some great documentation, which I would highly recommend reviewing. However, I have captured the key concepts below, including some input regarding our setup.

Workspace

A Linear Workspace is a container for all Issues, teams and other concepts relating to an individual business.

Teams

A Workspace can have one or many Teams, including nested Teams. Teams typically represent groups of people who work together frequently.
In our instance, a Team is a Product Team. For example, the “Web” Product Team. In a perfect world, we would have root Teams as “Product Groups” and an individual nested Team for each specific “Product Team”. For example, Customer (Product Group) > Web (Product Team). However, the concept of a nested Team is new and has some limitations (you cannot import to a nested Team). Teams can be individually configured or inherit standard settings from a root Team.

Issues

The most basic concept in Linear is the Issue. Most other concepts in Linear are either associated with Issues, or designed to manage Issues.
An Issue represents a task described in simple language. Each Issue must belong to a single Team, and have a unique identifier, for example, “FARM-123”.
Issues are required to have a title and status, which makes the creation of an Issue very fast. To learn more about Issues, I would recommend reviewing the article “Write issues not user stories”.

Workflows

As Issues progress, they move through a Workflow, which is a group of ordered Issue statuses defined per team. This is where integrations with tools such as GitHub become critical, automating the progress based on the development itself, ensuring information is timely and accurate.

Projects

Projects group Issues towards a specific, time-bound deliverable, like launching a new feature. Projects have their own pages that include all Issues related to a Project, as well as details, graphs, etc. Projects can be shared across multiple teams.

Milestones

Milestones are a concept used to further organise Issues inside an individual Project. For each Project, Milestones represent meaningful stages of completion for that Project, including the associated Issues.

Cycles

Cycles are similar to sprints and prioritise a set of Issues during a specific period. Cycles are automated, with a start date and duration, which will be repeated based on the specified cadence. Cycles do not end in a release, meaning Issues not completed during a Cycle roll over to the next Cycle.

Initiatives

Initiatives help organise Projects and align them to business essential wins (business priorities). This enables high-level planning across multiple Projects and longer durations, whilst providing a simplified view for leadership to track holistic progress.

To get started with Linear, head over to their website and sign up for their “Free” plan.

Browser Use

Fri, 21 Feb 2025 00:00:00 -0600

In my article “RPA RIP”, I outlined my concerns regarding the overuse of traditional Robotic Process Automation (RPA). As an alternative, I positioned API-centric automation, combined with Generative AI to create “intelligent agents” (also known as Agentic AI).

As documented over a series of articles (the latest being “Agentic AI, my team has developed and scaled a multi-model, multi-modal Generative AI framework for business, which primarily aims to accelerate productivity by streamlining common tasks.

Last year, we launched a new feature known as “Workflows” which allows a user to declaratively (no code or specialist knowledge required) create chained events using pre-approved components. These events can be triggered in real time, promoting experimentation and continuous improvement through iteration.

These Workflows focus on API-centric interactions, either as an input or an output. For example, taking an input event from a user or signal to trigger an interaction with a system of record, such as SAP or Workday.

In recent months, we have enhanced this capability with “Deep Research”, leveraging reasoning models (e.g., OpenAI o3) to construct more complicated workflows that are dynamically compiled (not pre-determined scripts).

We believe these Agentic AI capabilities will trigger the next wave of innovation regarding the use of Generative AI, further increasing the overall return on investment.

However, not every system, especially in an enterprise business, is API-centric. Interacting and/or automating these legacy systems can be a challenge. This is where traditional RPA has historically proven valuable, as it can simulate user interactions via the user interface itself.

Unfortunately, this approach is very brittle as it relies upon pre-determined scripts that can not adapt to change.

Therefore, we have been testing “Browser Use”, which is an open-source project, backed by Y Combinator, that connects AI Agents to the browser. This opens the door to browser automation (no direct API required), without the need for pre-determined scripts.

In short, Browser Use accepts input from the user as a standard prompt and then leverages a language model to plan and execute the request in real time.

The video below is a short demonstration of Browser Use finding and approving a pull request within GitHub.

As part of this demonstration, Browser Use was not given any context regarding how to complete the task or insight into the GitHub website. It executed the user request by navigating the user interface organically, comparable to a human interaction.

The results of our testing have been surprisingly good, especially considering the relative immaturity of the project. However, it should be noted that the complexity of the task being requested and the specific language model being used can have a significant impact on the success of the process.

In addition, as the process is not pre-determined, the same input can result in different (inconsistent) results. Therefore, using these capabilities at scale in production would likely require robust grounding and thorough testing.

With that said, Browser Use does boast high levels of accuracy compared to other options, including “Operator” from OpenAI, which currently costs £200 per month to use.

Browser Use highlight the following features on their website. However, I would recommend reviewing their documentation for a more detailed description of the capabilities.

Vision + HTML Extraction

Combines visual understanding with HTML structure extraction for comprehensive web interaction.

Multi-tab Management

Automatically handles multiple browser tabs for complex workflows and parallel processing.

Element Tracking

Extracts clicked elements XPaths and repeats exact LLM actions for consistent automation.

Custom Actions

Add your own actions like saving files, database operations, notifications, or human input handling.

Self-correcting

Intelligent error handling and automatic recovery for robust automation workflows.

Any LLM Support

Compatible with all LangChain LLMs including GPT-4, Claude 3, and Llama 2.

As Agentic AI gains momentum, I expect to see a lot of browser automation capabilities appear on the market. However, I believe Browser Use has emerged as an early favourite, achieving 40k stars on GitHub in three months. They also have a strong community (open source) and great backing from established investors (Y Combinator).

Overall, the best form of automation is still programmatically defined, using standard APIs. However, knowing that this is not viable for all scenarios, especially in enterprise businesses, I am excited to see modern alternatives becoming available.

Microsoft Majorana 1

Wed, 19 Feb 2025 00:00:00 -0600

In December, I published the article “Google Willow”, covering Google’s recent Quantum Computing announcement.

I started researching Quantum Computing in 2015, supported by an engagement with IBM to understand the potential for life sciences, specifically drug discovery.

Over the years, I have written a few articles covering Quantum Computing, which provide a good foundation for anyone interested in learning more about the topic.

Within the mainstream media, Quantum Computing had gone a little quiet.

Therefore, the Google Willow announcement was a welcome update and a positive sign that progress is still being made. As a reminder, Google Willow demonstrated the ability to reduce errors exponentially as quantum processors scale up using more qubits. This is a key challenge in quantum error correction.

Earlier today, Microsoft announced Majorana 1, positioned as the first quantum chip powered by a new Topological Core architecture.

The video below from Microsoft summarises the announcement.

A topological qubit is a special kind of qubit, which has historically proven more difficult to create than regular qubits.

In theory, a topological qubit is more resilient to decoherence than regular qubits, providing a robust foundation. If proven scalable, this improved reliability could deliver a significant advantage over regular qubits.

With that said, although this is a milestone for topological quantum computing, it is still a long way behind non-topological approaches.

To help understand the impact of this announcement, I would recommend reading the reaction from Scott Aaronson (Acclaimed Theoretical Computer Scientist).

As with Artificial Intelligence, it would appear Google and Microsoft are taking a leadership position in quantum computing. It will be fascinating to see which company (and approach) will rule supreme.

AHNTIE 2025

Sat, 15 Feb 2025 00:00:00 -0600

Earlier this week, I attended the Animal Health, Nutrition and Technology Innovation Europe (AHNTIE) conference in London.

This multi-day event includes over 800 attendees from across the animal health industry, covering enterprise businesses, service providers, investors, academics, and over 250 startups.

My company was the headline sponsor, with our CEO Jeff Simmons delivering the keynote as part of a fireside chat.

As a technologist, it was interesting to see the impact of Artificial Intelligence (AI), which was everywhere. Every speaker and conversation referenced AI, with opportunities covering the entire value chain from drug discovery to customer engagement.

Although exciting, it was evident that the animal health industry is still very immature in this area, with clear gaps in understanding and/or a lack of clarity regarding the specifics of how AI could/should be applied.

With that said, it is clear the “mainstream media buzz” has captured the imagination of the industry, which certainly presents an opportunity for technologists to build on this momentum.

Due to the intensity and energy of the event, it is not possible to provide a comprehensive overview. Therefore, I have chosen to summarise a few key messages that resonated strongly with me.

Key Messages

The animal health industry is very robust and resilient. It has the potential for sustained year-on-year growth, with a low risk of a “crash”, as seen by other more volatile industries (e.g., DeepSeek and NVIDIA).
In 2013, standalone animal health companies served 37% of the industry. In 2025, the market has shifted dramatically, with standalone animal health companies representing 72%.

We continue to see the humanisation of pets, where animals started outside, migrating inside, and are now in our beds. Pets are increasingly treated as a member of the family.
Unlike human health, across pets and farms, customer loyalty is very strong, with examples of 50-year-old products continuing to sell.
Pets are living longer, specifically dogs and cats. This is great, but it creates new challenges for old-age care.
The veterinary industry is under increasing pressure and overworked (commonly 3 full-time employees per vet practice). Veterinarians desire new ways of working to meet demand. Excitement that Artificial Intelligence (AI) could increase efficiency and effectiveness, which some Veterinarians consider a higher priority than new drug products.
Success in the animal health industry is driven by science and innovation. Human health breakthroughs remain a good indicator for future animal health innovation. For example, the adoption and impact of Monoclonal Antibodies (mAbs) in human health is approximately 25%, compared to 5% in animal health. This highlights a clear growth opportunity with a proven path to success. With animal health, Monoclonal Antibodies (mAbs) currently represent $1.2B of the $41B core market.

The use of animal protein continues to be scrutinised. However, animal protein demand continues to grow.
66% of individuals under 30 care about animal welfare and environmental impact. Therefore, products focused on sustainability remain highly relevant, with climate-neutral farming an essential goal.
Similar to other industries, we are seeing an attack on science and innovation, where facts are distorted and/or ignored by small, but vocal social groups. Animal health is a science-based industry. Therefore, the industry must stay unified to educate, promote science and protect innovation.
Those who raise and care for animals are naturally passionate and curious. This passion must be considered when entering the market, with a focus on feeling connected and/or being part of something bigger.
Pet owner spending is increasing, with growth across “feel-good” items, such as toys, treats and technology. In the US, the younger the pet owner, the more they spend (which is great for the industry).

Since 2020, the cost of dog ownership in the US has been rising, with vet services +56%, medications +34%, food and treats +46% and grooming/kennels +52%. The cost of cat ownership in the US has not increased at the same rate but is still +17% overall.
Concerningly, the use of vet services is trending down (approximately 1.9% in the US). With the average time between visits being 86 days or 112 days excluding medication collection.
$100M products within animal health are classified as a “blockbuster”, with 15 blockbusters currently on the market. We are starting to see the first $1B brands (a family of complimentary products).
Science is the ultimate disruptor within animal health, with tremendous opportunities to leverage computational science and Artificial Intelligence (AI) to accelerate the drug discovery process. This trend increases the importance of gaining access to expertise and since 2020 datasets, such as genomics data.

Conclusion

Overall, it was great to be part of the event and to interact with this passionate community!

As I look to the future, I believe the Stonehaven Cozmix Group summarised the near-term opportunities well, covering the following areas across pet and farm.

Pet:

Monoclonal Antibodies (mAbs)
Adeno-associated Virus Gene Therapy (AAV)
CAR-T Cell Therapy

Farm:

One Health (Epidemics, Antimicrobial Resistance, Antiparasitic Resistance)
Sustainability (Microbiome on the Rise)
Scarcity of Skilled Labour

The animal health industry tends to be slower moving, with a focus on micro-innovation vs. revolutions. However, there is a growing appreciation for how technology can be applied across the value chain to deliver compelling outcomes, covering drug discovery, customer engagement and operational effectiveness.

As a result, it is an exciting time to be a technologist working in the animal health industry.

I am excited to see the next wave of breakthrough innovation hitting the market, delivering value to those who raise and care for animals around the world.

Product Operating Model Transition

Mon, 10 Feb 2025 00:00:00 -0600

In my previous three articles, I outlined the Product Operating Model, as defined by the Silicon Valley Product Group (SVPG).

The Silicon Valley Product Group (SVPG) has published three books on the Product Operating Model. Specificlly, INSPIRED, EMPOWERED, and TRANSFORMED.

These books are an incredible resource. However, the case studies referenced focus on digital-first / technology-first companies, specifically Netflix, Spotify, etc. These are certainly fascinating examples but do not represent the majority of businesses.

For example, Spotify is a 19-year-old company born in the Internet era. It predominantly produces digital (software) products and has minimal regulatory oversight.

I have spent my career working at two life science companies, one that is 150 years old and another that is 70 years old. These companies have a complex value chain, predominately producing drug products (not digital products), and operating within a highly regulated industry with strict compliance controls.

In addition, the years of heritage and legacy create inertia, incurring human and technical debt that can act as a barrier to change.

Therefore, the comparisons with Netflix and/or Spotify lack credibility.

With this in mind, it would be easy to dismiss the Product Operating Model as not relevant for any business that is not “digital native”. However, in my opinion, the product-first principles are highly applicable, helping any business (regardless of industry) to focus on what matters most, the outcome and customer value.

In 2020, we started our journey towards the Product Operating Model. Therefore, I thought I would share more detail regarding our transition approach.

Product Operating Model Transition

In a perfect world, any transition to a Product Operating Model would be sponsored by the CEO and include active engagement from the Executive Committee.

Unfortunately, convincing a CEO to transition the business operating model is no easy feat, usually requiring an intervention that forces the change. For example, a change in leadership or a market disruption.

As a result, our approach was to initiate the change from IT. There is logic starting with IT, acknowledging that almost every industry and business is increasingly reliant on technology and data to differentiate and drive value.

It should be noted, that this approach is most viable when IT is already established as a trusted partner with a focus on value generation. In some businesses, IT is exclusively positioned as a support function, which would likely act as a barrier when attempting to transition to the Product Operating Model.

A core aspect of the Product Operating Model is the product-first principles. These twenty principles place an emphasis on value, viability, usability, and feasibility, which are highly relevant to ensure a successful outcome.

The Product Operating Model and the product-first principles are best positioned when there is a “greenfield” problem to be solved. This means there is no predetermined solution defined and the outcome is within the control of the product team (minimal restrictive dependencies). For example, within my industry, a big “problem” is how to accelerate the drug discovery process.

In this scenario, a Product Team, consisting of a Product Manager, Product Designer and Technical Lead (backed by engineers) are empowered to solve the problem, starting with Product Discovery, before transitioning to Product Delivery. This team is durable and therefore persists through the duration of the product lifecycle.

This scenario follows the exact approach as defined by the Product Operating Model. I have also written about the specific roles/responsibilities in the article “Product Operating Model Roles”.

The challenge is that not every problem is “greenfield”. For example, some problems have predetermined solutions (maybe due to existing investments) or the product team do not have complete control to shape/influence the outcome (maybe due to a regulatory constraint).

In addition, recognising the wider role of IT, there are many scenarios where the problem has already been solved and therefore, there is no value in discovering a new solution. For example, commodity capabilities (e.g., Hosting, Telephony) and/or highly industrialised processes (e.g., Human Resource Planning).

These scenarios are classified as “enablement”, where a product team is still assigned, however, the primary mission of the team is to deliver a solution that is scalable, secure, reliable and efficient. As a result, the Product Designer, Technical Lead and Engineers are dynamically assigned (not durable), based on corporate priorities and/or capacity. This helps to manage total headcount and capacity planning.

The Silicon Valley Product Group (SVPG) refer to this type of team as “maintenance”. However, in my opinion, the word “maintenance” undervalues the teams, as their contributions are critical, commonly providing the foundation for other product teams to build upon. Hence, I prefer the word “enablement”, which more accurately describes the value proposition.

Regardless of the team outcome, they all follow the product-first principles, promoting consistent solutions. This approach recognises that every team has a problem to solve, customers to understand/delight, and risks and constraints to manage.

The key is to avoid bifurcation (bi-modal) operation, which can unintentionally relegate the value of certain teams, whilst also creating confusion and friction.

With this context, IT is organised into product teams, logically grouped under Product Leaders (that focus on people enablement and provide strategic context). Many of these product teams are durable (persistent) and focused on the “big problems”, aligned with the business strategy. The product team topology will be different for every business and it is not fixed, dynamically evolving based on the business priorities.

It is my belief that this model, adapted from the Product Operating Model, following the product-first principles, provides a versatile foundation that promotes continuous innovation and delivery excellence, helping to ensure that every solution achieves the desired outcome.

PC Build Performance

Sat, 25 Jan 2025 00:00:00 -0600

Last week, I purchased the components for my custom-built PC upgrade.

This was my first major upgrade since 2021, in preparation for the launch of the NVIDIA GeForce RTX 5090.

Unfortunately, the NVIDIA GeForce RTX 50 series has had a very disappointing launch, with restrained availability worldwide, resulting in aggressive scalping that has pushed the prices up significantly. For example, the RRP of the NVIDIA GeForce RTX 5090 Founders Edition is £1939 but is currently being sold on eBay for £3,500 and beyond.

In addition, those who did receive an NVIDIA GeForce RTX 5090 are facing some early adopter challenges, such as driver errors and (more concerningly) issues with the 12VHPWR power connector overheating and melting.

This entire situation is frustrating and could have been easily avoided. I recognise that the consumer graphics card market represents only a tiny part of the NVIDIA business. However, considering their heritage, I wish they would treat their loyal customers with more respect.

I refuse to pay over the RRP for a graphics card that is arguably already overpriced for the performance. Therefore, I will continue to use my NVIDIA GeForce RTX 3090 Founders Edition until the market stabilises.

The specification of my current (soon-to-be old) PC can be found below.

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
NVIDIA GeForce RTX 3090 FE 24GB GDDR6x
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Fractal Torrent Compact
Samsung G95NC Odyssey Neo G9 57-inch Monitor

I purchased the following new components, retaining my existing graphics card, case, monitor and peripherals.

MSI MAG X870 Tomahawk WiFi
AMD Ryzen 9 9950X 4.3GHz Base / 5.7GHz Boost (16C/32T)
Noctua NH-D15 G2 LBC CPU Cooler
64GB Corsair Vengeance EXPO DDR5 PC5-48000C30 6000MHz RAM
2TB Crucial T705 M.2 PCI-e Gen5 NVM-e SSD
Corsair HX1200i 1200W ‘80 Plus Platinum’ ATX PSU

The full details of the upgrade, including the rationale for each component, can be found in my previous article “GeForce RTX 5090 Build”.

I have now built the PC and can share some initial performance results. As previously stated, I am still using my old NVIDIA GeForce RTX 3090. Therefore, performance benefits targetting graphics-intensive workloads will be limited.

Build

The build process itself was very straightforward, with no notable issues. Over the years, I have built countless computers, but this was arguably the simplest ever.

The only point of “tension” was the application of the thermal compound on the processor. I use Noctua NT-H2, which has proven to be highly performant and does not require any special preparation (no break-in or burn-in time).

I follow the Noctua recommended approach for thermal compound application, specifically one larger dot in the centre, with four smaller dots at each edge. Recognising the unorthodox shape of the AMD Ryzen 9950X heat spreader, I also used the Noctua NA-TPG1 thermal paste guard.

Finally, I tested the system prior to completing my cable tidy process (a mistake I have made in the past).

With the hardware installed, I reset the BIOS to factory default, downloaded the latest Windows 11 ISO from Microsoft and triggered the setup process.

Frustratingly, the Windows 11 ISO did not recognise the network devices (Ethernet or WiFi) during the setup. Due to the ridiculous and infuriating requirement for an Internet connection, I had to perform the bypass procedure.

The bypass procedure is achieved by clicking “Shift + F10” during the Windows 11 setup process (which launches the command prompt) and typing ‘oobe\BypassNRO’. At this point, the setup process will restart, providing an option to skip network connectivity.

Over the past few years, I have become increasingly disappointed by Windows as an operating system, which has quickly become a vehicle for Microsoft to publish advertisements, attempting to sell their services such as Microsoft 365, Copilot and GamePass.

I would love to switch to Linux full-time. However, unfortunately, there are still too many blockers that would impact my common workloads (e.g., Microsoft Office Suite, Games with Anti-Cheat, etc.)

With Windows 11 finally installed and updated, I confirmed system stability, before updating and configuring the BIOS for optimal settings. I enabled the AMD Extended Profiles for Overclocking (AMD EXPO) and AMD Precision Boost.

I plan to overclock the system using a combination of AMD Precision Boost and AMD Curve Optimiser. However, I prefer to test “default” performance and reliability before configuring more advanced settings.

Performance

The table below provides the results from a series of tests, highlighting the performance improvements over my old AMD Ryzen 9 3950X PC.

As expected, the jump from the AMD Ryzen 9 3950X to the AMD Ryzen 9 9950X is significant, covering single-threaded and multi-threaded workloads.

The memory and storage results were also impressive, supported by the jump to DDR5 and the ridiculous performance of the Crucial T705 M.2 PCI-e Gen5 NVM-e SSD, delivering 14,500MB/s for sequential reads and 12,700MB/s for sequential writes.

It should be noted, that these performance results (specifically the storage results) do not necessarily represent “real-world” increases.

On average, at default settings, the upgrade has delivered a 93% performance increase for CPU/Memory focused tasks and 139% for storage focused tasks, which was above my expectations.

In a future article, I will share more details regarding my configuration and start to explore AMD Precision Boost and AMD Curve Optimiser to unlock additional performance improvements.

GeForce RTX 5090 Build

Sat, 18 Jan 2025 00:00:00 -0600

The last time I completed a major upgrade of my primary PC was in 2021, when I switched to an AMD 3950X from an Intel Core i7-6700K.

I documented the build process and subsequent upgrades in the following four articles.

The full specification of my current (soon-to-be old) PC can be found below:

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
NVIDIA GeForce RTX 3090 FE 24GB GDDR6X
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Fractal Torrent Compact

This PC has served me extremely well over the past four years, delivering great performance across a wide range of workloads.

The use of high-performance parts also allowed me to skip several generational upgrades, specifically the AMD Ryzen 5000 and 7000 series (Zen 4), as well as the NVIDIA GeForce RTX 40 series.

However, at CES 2025, NVIDIA announced the highly anticipated GeForce RTX 50 series of consumer graphics cards. Specifically, the GeForce RTX 5090, based on the NVIDIA Blackwell architecture.

Although not yet available the specification of the GeForce RTX 5090 guarantees a significant performance increase over my GeForce RTX 3090. This is thanks to the increase in CUDA cores, high clock speeds, and faster memory, as well as the introduction of new features such as 4th Generation RT cores, 5th Generation Tensor cores and DLSS4.

The combination of improved specification and new features, was enough to spark my interest, resulting in the decision to upgrade my entire system.

This article outlines the details of the upgrade, including the rationale behind the decisions.

Working in IT, I utilise my PC for a wide range of different workloads, including productivity, collaboration, software development, local artificial intelligence, photo editing, video editing, video effects, virtual labs, gaming and game development.

Similar to my last build, these requirements demand high-performance components and my goal is to design a balanced system that will perform well in all scenarios.

As a result, I have selected the following hardware. Any hardware not listed (e.g., case) will be reused from my current build.

MSI MAG X870 Tomahawk WiFi
AMD Ryzen 9 9950X 4.3GHz Base / 5.7GHz Boost (16C/32T)
Noctua NH-D15 G2 LBC CPU Cooler
64GB Corsair Vengeance EXPO DDR5 PC5-48000C30 6000MHz RAM
2TB Crucial T705 M.2 PCI-e Gen5 NVM-e SSD
Corsair HX1200i 1200W ‘80 Plus Platinum’ ATX PSU
NVIDIA GeForce RTX 5090 FE 32GB GDDR7

Motherboard

My previous MSI MAG X570 Tomahawk WiFi had served me well, delivering reliability and high performance, even when overclocked. I was also impressed by the price/feature ratio, build quality, and BIOS support, which all compare favourably to the competition (e.g., Asus, etc.).

The latest chipsets from AMD are the X870 and X870E. On paper, these chipsets are a little underwhelming, as they are iterations of the previous B650E and X670 chipsets, with the main difference being the addition of USB 4 (40Gb/s). With that said, iteration is not always bad, hopefully allowing the AMD product team to focus on refinement.

Unfortunately, since I last purchased a motherboard, the prices have increased significantly, with X870 motherboards ranging from £200 to £600.

I selected the MSI MAG X870 Tomahawk WiFi, which I believe delivers the best price/feature ratio.

The MSI MAG X870 Tomahawk WiFi is an X870 chipset, not an X870E. The X870E is a dual-chip architecture (incorporating two Promontory21 chipsets), which delivers additional PCI-e lanes and USB I/O. However, the dual-chip architecture also adds complexity, heat and cost. Therefore, unless you have a specific purpose for the PCI-e lanes and/or USB I/O, the X870E is likely not worth the additional investment.

Outlined below are the key features of the MSI MAG X870 Tomahawk WiFi motherboard.

AMD X870 Chipset
14+2+1 Duet Rail Power System (14-phase, 80A Power Stages)
AM5 Socket
4x DDR5 UDIMM
1x PCIe Gen5 x16 (CPU)
1x PCIe Gen4 x4 (Chipset)
1x PCIe Gen3 x1 (Chipset)
2x M.2 Gen5 x4 (CPU)
1x M.2 Gen4 x4 (Chipset)
1x M.2 Gen4 x2 (Chipset)
4x SATA (6Gb/s)
2x USB 4 (40Gb/s)
1x USB 3.2 (20Gb/s)
3x USB 3.2 (10Gb/s)
7x USB 3.2 (5Gb/s)
8x USB 2.0
1x 5Gb/s LAN
WiFi 7
Bluetooth 5.4

Needless to say, that is a lot of connectivity, which can be confusing. Therefore, I would recommend reviewing the motherboard manual, specifically details regarding the bifurcation of the PCI-e lanes, which can limit/disable certain features based on the hardware installed (e.g., USB 4).

To see how the MSI MAG X870 Tomahawk WiFi performs against the competition, I recommend reviewing the TechSpot X870/X870E Motherboard Roundup.

The image below from TechSpot highlights the testing results of the MSI MAG X870 Tomahawk WiFi. As stated in their article, “performance-wise, the Tomahawk excelled, averaging a clock speed of 5,110 MHz for a score of 2,306 points, one of the highest scores in our testing.”

The image below shows the integrated rear I/O panel of the MSI MAG X870 Tomahawk WiFi. Note, the accessible Clear/Flash CMOS buttons, which can be very useful when tuning or overclocking a system.

At £279.98 (Scan) the MSI MAG X870 Tomahawk WiFi is not the cheapest X870 motherboard but based on my experience with MSI, will deliver a high quality, reliable foundation, with critically reviewed features and performance that rivals more expensive motherboards from Asus, Gigabyte, etc.

Processor and Cooler

If I were building a PC specifically for gaming, I would select the new AMD Ryzen 9800X3D, which thanks to the second-generation AMD 3D V-Cache, delivers best-in-class gaming performance.

With that said, the Ryzen 9800X3D is an eight-core, sixteen-thread processor, with a maximum boost clock of 5.2 GHz. Therefore, although the Ryzen 9800X3D is the gaming champion, it falls behind in productivity or specialist tasks (e.g., development, video editing) compared to processors with a higher core count and/or frequency.

As a result, I have selected the AMD Ryzen 9 9950X, which is a sixteen-core, thirty-two-thread processor, with a maximum boost clock of 5.7 GHz. Although the AMD Ryzen 9 9950X received mixed critical reviews, this was mostly driven by inflated expectations associated with the Zen 5 architecture, which many had unrealistically hoped would deliver significant (> 30%) performance benefits over Zen 4.

In my case, the AMD Ryzen 9 9950X delivers a massive performance increase over my current AMD Ryzen 9 3950X, which is based on the Zen 3 architecture. In addition, it successfully balances performance across a wide range of workloads, unlike the AMD Ryzen 9800X3D, which would compromise productivity.

Arguably, I should wait for the recently announced AMD Ryzen 9950X3D to be released. However, I am not convinced by the benefits of 3D V-Cache for processors with dual Core Chiplet Dies (CCDs), specifically the AMD Ryzen 9900X/9950X.

These processors only have a 3D V-Cache on one die, which requires the software (usually Microsoft Windows) to accurately schedule tasks to see the desired performance benefits.

This additional complexity has been proven to cause issues, requiring more “human effort” to maximise performance across workloads. In addition, 3D V-Cache commands a price premium of approximately £200, making it a significant investment for questionable value.

Finally, following recent discounts, the AMD Ryzen 9 9950X is currently the same price as the AMD Ryzen 9800X3D (approximately £529.99), making it cost-competitive.

To manage the thermals of the AMD Ryzen 9 9950X, I have selected the Noctua NH-D15 G2 LBC CPU Cooler.

This will be my third Noctua NH-D15 and I have always been impressed with the build quality and thermal results. The recent release of the NH-D15 G2 in theory delivers even better results, although I suspect the real-world difference will be minimal.

Either way, outside of custom water cooling (which I have no desire to implement), I continue to put my trust in Noctua.

I fully expect to be able to leverage AMD Precision Boost Overdrive (PBO) and Curve Optimiser to deliver a modest 24x7 overclock, whilst still maintaining acceptable thermals and noise.

Memory

I do not have extreme overclocking expectations. However, I do plan to leverage AMD Extended Profiles for Overclocking (EXPO). To ensure system stability, AMD recommends a maximum of two DIMMs.

As a result, I have selected 64GB (2x32GB) Corsair Vengeance DD5 PC5-48000C30 RAM, which is EXPO (AMD) configured, operating at 6000MHz with 30-36-36-76 timings.

This is not the highest frequency memory, but my research indicates that 6000MHz/6400MHz with tight timings (< 32 CAS Latency) is the sweet spot for optimal performance with AMD Ryzen 9000 series processors.

Higher frequency memory commonly comes with less aggressive timings, which can offset the performance benefits. It also places greater stress on the system, which can impact stability or require a lot of additional tuning.

64GB offers decent headroom, but I would have preferred 96GB (2x48GB). Specialist workloads such as development, local artificial intelligence, and virtual labs can easily consume 60GB+.

Unfortunately, 96GB options are currently limited and costly. Thankfully, memory is the easiest component to upgrade. Therefore, I will keep an eye on the market and potentially upgrade in the future.

Storage

The MSI MAG X570 Tomahawk WiFi has two M.2 PCI-e Gen5 x4 slots, both of which are connected to the processor. I have decided to occupy one slot with the Crucial T705 M.2 PCI-e Gen5 NVM-e SSD.

The Crucial T705 is one of the highest-performing consumer-grade drives available today, with the 2TB version capable of reaching up to 14,500MB/s for sequential reads and 12,700MB/s for sequential writes, as well as up to 1,550K random read IOPS and 1,800K random write IOPS.

It uses the Phison E26 SSD controller (based on the Max14um platform) and Micron 232-Layer TLC NAND flash, similar to what is found on competing products, such as the Corsair MP700 Pro SE. This combination delivers an impressive bus speed of 2,400MT/s, which unlocks the incredible sequential read performance.

In day-to-day scenarios, the impact of this cutting-edge performance is likely negligible. However, for certain storage-heavy workloads, the additional performance offered by the Crucial T705 will reduce the risk of bottlenecks. The drive is also optimised for Microsoft DirectStorage, making it future-proof for gaming.

Power Supply Unit (PSU)

I have selected the Corsair HX1200i 1200W ‘80 Plus Platinum’ ATX PSU, which is ATX 3.1 and PCI-e 5.1 certified, delivering 1200W via a single rail. It also comes equipped with three EPS12Va connectors and a PCI-e 5.1 12V-2x6 GPU cable.

The NVIDIA GeForce RTX 5090 specification states a 575W Thermal Design Power (TDP), which is up from my current 350W of the NVIDIA GeForce RTX 3090.

This extreme power requirement demands a high-performance, high-quality power supply unit to ensure safe/reliable power delivery.

When calculating my maximum system wattage, I assumed the following worst-case scenario.

Processor Overclocked (25%)
Graphics Card Overclocked (10%)
2x SSD
2x RAM
5x Fans
USB 3.2 Gen2 Power Transmission

These assumptions resulted in a maximum system power requirement of approximately 1060W.

NVIDIA officially recommend a minimum 1000W PSU for the GeForce RTX 5090. However, I would always recommend selecting a PSU with plenty of headroom.

Conclusion

With my research complete, I have procured the core components and now await the official launch of the NVIDIA GeForce RTX 5090.

In the interim, I plan to build and tune the system using my current NVIDIA GeForce RTX 3090.

As with previous builds, I plan to document the process, including my 24x7 settings and performance results.

Google AI Business Trends

Fri, 17 Jan 2025 00:00:00 -0600

Google recently published their AI Business Trends 2025 presentation.

Google Cloud AI Business Trends 2025

The document highlights five key trends.

Multimodal AI - The Integration of Diverse Data Sources
AI Agents - Managing Complex Workflows
Assistive Search - The Next Frontier for Knowledge Work
AI Customer Experience - Technology is Invisible
Security - Advanced Techniques to Identify and Combat Threats

The document targets business leaders and therefore does not provide technical details. However, it does a good job of summarising the key themes, backed by relevant data points and references.

I would specifically draw your attention to slide 20 (Employee Agents), where Google referenced the great work being done by my team.

If you want to learn more, I have documented our journey over a series of blog posts.

Congratulations to the team for the public recognition!

RPA RIP

Mon, 06 Jan 2025 00:00:00 -0600

I am a strong advocate for business automation, helping to streamline common tasks to ensure human intelligence can be applied to higher-value opportunities, including areas that require curiosity, intuition, etc.

However, it is important to note that not all automation is created equal.

Over the past decade, I have been vocal about my concerns regarding the overuse of traditional Robotic Process Automation (RPA), which frequently conflicts with recommendations from service integrators and consultants.

My concern is simple, traditional RPA is commonly built on 20+ year-old technology, reliant on screen scraping (simulated user interface) techniques. In many respects, it is similar in concept to Selenium, the browser-based regression automation suite, first developed in 2004.

The value of RPA is the ability to replace human interactions (at the user interface level), allowing businesses to reduce or redeploy headcount. This delivers a compelling short-term ROI, which is why it is favoured by service integrators and consultants, as they can demonstrate “quick wins” with no accountability for future ramifications.

Unfortunately, this approach is very short-sighted, it essentially perpetuates legacy business processes, which can have disastrous medium/long-term ROI implications and impact future innovation.

This issue is compounded by the fact that traditional RPA is very “brittle” as it is often coupled to the user interface of the application it is automating. Therefore, any change to the application immediately breaks the automation (there is no intelligence).

To make matters worse, traditional RPA replicates the user and therefore requires user-level authentication, permissions and licensing, which impacts security and operational costs.

Finally, traditional RPA does not scale effectively. Each automation (bot) commonly requires an individual Windows-based virtual machine (or client), including access to the software that needs to be automated. This can quickly become complex and costly to support, further impacting the medium/long-term ROI.

Thankfully, there is another option, leveraging modern techniques and architecture to deliver API-centric automation that is loosely coupled, delivering flexibility, security, scale and cost-effectiveness. It also presents the opportunity to re-think the business process itself, which in my opinion, should always be the starting point before any automation is considered.

The one downside is the requirement to have access to viable API endpoints. In modern software, these are usually natively available. However, legacy software may require an additional intervention, adding complexity to the initial implementation.

With the introduction of Generative AI, including projects such as LangChain, we started to see evidence that API-centric automation could be combined with Generative AI to create “intelligent agents”.

These agents, with specific grounding, have the ability to take action and learn, whilst remaining loosely coupled to protect flexibility, as well as maintaining modern security standards.

In November, the venture capital firm a16z released the blog post “RIP to RPA: The Rise of Intelligent Automation”, which provides a great overview of this emerging trend.

In addition, a16z published a short video discussing the topic.

I agreed with their premise and am pleased to see others promoting modern approaches to automation, whilst highlighting the drawbacks associated with traditional RPA.

This year, I expect “Agents” and “Agentic AI” to enter the mainstream, likely positioned as the next evolution of Generative AI.

As a result of this market shift, I expect many traditional RPA vendors to “rebrand”, distancing themselves from “RPA” terminology, and replacing it with “Agentic AI”. This approach is “putting lipstick on a pig”. Therefore, I recommend a thorough investigation of the underlying architecture (looking beyond the sales pitch and marketing) before any investment.

As an alternative, I recommend new engagements start with the leaders/innovators in Generative AI and/or open-source alternatives. For example, Project Mariner from Google DeepMind and Operator from OpenAI.

My team has been experimenting with Agentic AI for many months. We intend to deliver a robust AI automation framework for our business, with dynamic workflows that can support a wide range of common and complex business tasks.

Agentic AI

Mon, 16 Dec 2024 00:00:00 -0600

The work was recently recognised by Google Cloud CEO Thomas Kurian during the “Gemini at Work” global digital event opening keynote.

I have documented the journey over a series of blog posts.

As we close the year, our Generative AI framework has reached another significant milestone (see the dashboard screenshot below).

Specifically, we have passed 7,000 unique users. This is important, as my company has approximately 9,000 employees (excluding contractors), some of which do not have access to IT services. At the start of the journey, we predicted that 7,000 was the viable audience for Generative AI, meaning we have reached peak adoption.

This has delivered an estimated return on investment of $2,419,808, predominately through efficiency gains and cost avoidance. I shared more details regarding specific success stories in the article “Generative AI - Value”.

A major part of this success has been our marketplace, which allows any employee to securely create and share a chatbot (known as a custom persona), including specific business context and the ability to ingest data (unstructured and structured) from multiple source systems.

We now have over 1000 unique chatbots, all built on the same underlying cost-effective architecture. These chatbots support individuals and teams, covering the entire value chain from Research & Development through to Commercial Sales and Marketing.

As highlighted by the slide below, this private, secure and compliant foundation leads us to our next area of opportunity, specifically action and learning.

Currently, our Generative AI framework responds to prompts and executes predefined tasks. In the future, we expect it to make decisions, plan and execute actions, and even learn. The industry has coined the term “Agentic AI” to describe this evolution.

We have already launched our first version of this capability, which we call “Workflows”.

In short, workflows allow any user to declaratively (no code or specialist knowledge required) create chained events using pre-approved components. These events can be triggered in real-time, promoting experimentation and continuous improvement through iteration.

For example, a workflow could accept input from an external system or data source via an API. This input could be ingested and processed by multiple AI models, including self-verification and/or summarisation of the output. The workflow could then dynamically trigger a downstream action based on the input, including an API call to an external system or data source.

In our early testing, we have been able to create some very interesting workflows, pulling data from systems such as SAP, and analysing the data using a reasoning model, such as OpenAI o1 and Google Gemini 2.0, producing a structured output that includes text and images.

We believe this capability will trigger the next wave of innovation regarding the use of Generative AI, further increasing the overall return on investment. It also presents a far more modern, decoupled and less brittle approach to process automation, where the workflows are programmatically defined (reusable) and can dynamically adapt based on an ever-evolving landscape.

We are not alone in this thinking, with Google demonstrating similar capabilities within their AI Studio and as part of Project Mariner.

Bowers and Wilkins Px8

Sun, 15 Dec 2024 00:00:00 -0600

The Bowers & Wilkins Px8 are premium active noise cancelling (ANC) over-ear headphones.

I have used Bowers & Wilkins products over the past decade, including multiple headphones and the excellent Mini Theater, connected to a Sony STR-DN1050 AV Receiver.

The UK-based Bowers & Wilkins brand has always resonated with me, specifically their rich heritage of high-performance bespoke sound hardware and their elegant design language that expresses their unique, premium style.

Most importantly, the sound produced by their products has always exceeded my expectations, offering clarity and punch that is faithful to the source material. In my experience, this balance is with consumer-centric sound products, often failing to deliver clarity or exaggerating specific ranges, such as bass.

I have been interested in the Bowers & Wilkins Px8 headphones for some time. However, with a recommended retail price of £599, I was unwilling to pull the trigger. Thankfully, during key sales periods, they can be found for around £399 and I managed to procure them for £290.19 (51.55% discount) during a recent sale on Amazon.de.

At this price point, the Bowers & Wilkins Px8 offer great value, especially when compared against the competition.

Bang & Olufsen Beoplay H95 (RRP £879)
Apple AirPods Max (RRP £499)
Bose QuietComfort Ultra Headphones (RRP £449.95)
Sony WH-1000XM5 (RRP £379)

The Bowers & Wilkins Px8 headphones exude quality. The premium materials, including cast aluminium arms, diamond-cut detailing and Nappa leather trim, look and feel amazing. The fit is also exceptional, delivering a reassuring and comfortable clamping force that (for me) feels more secure than the Apple AirPods Max (they are also approximately 64.8g lighter).

Highlighting the attention to detail, Bowers & Wilkins even angle the drivers, so that they are parallel to your ears to help with timing and precision in the sound (a nice touch).

Regarding the drivers, Bowers & Wilkins include new carbon cone 40mm drive units, which are lighter and claim to be more rigid, reducing distortion and improving transparency. This is supported by high-resolution 24-bit digital signal processing (DSP).

In summary, I would highlight the following positives:

Exceptional Sound
Premium Design and Materials
Refined Build Quality
Physical Buttons (Easy to Use)
Excellent Battery Life (30 Hours)
High-Quality Codec Support (aptX Adaptive, aptX and aptX HD)
Versatile Connectivity (Bluetooth 5.2, USB-C and 3.5mm Headphone Jack)

Without question, the main positive is the incredible sound performance, which is the best I have experienced with active noise cancelling (ANC) over-ear headphones. This is also the consensus from many audio critics, for example, What H-Fi.

The active noise cancelling and transparency modes are both excellent, but arguably not quite as good as the Apple AirPods Max or Sony WH-1000XM5. However, if your use case is mostly home or office (not travelling), the Bowers & Wilkins Px8 will certainly deliver.

One critique, the Bowers & Wilkins Px8 headphones can not be used passively, meaning they need to be “on” for them to receive sound. This is not uncommon for active noise cancelling headphones and on the surface is a minor concern.

However, this negative is compounded by the fact that the internal battery cannot be replaced easily. As with any battery, it will degrade over time. When this happens the headphones may become unusable. In a perfect world, Bowers & Wilkins would make the battery user replaceable, or at minimum, offer a simple low-cost replacement service. At this time, neither option is viable.

This critique would not stop me from recommending the Bowers & Wilkins Px8 headphones. However, it would give me pause at the recommended retail price of £599, knowing they will have a predetermined usable life span.

Model Context Protocol

Wed, 11 Dec 2024 00:00:00 -0600

Last month, Anthorpic introduced a new open-source standard for connecting AI assistants, known as Model Context Protocol (MCP).

The the blog post “Introducing the Model Context Protocol”, Anthorpic state:

*“Even the most sophisticated models are constrained by their isolation from data—trapped behind information silos and legacy systems. Every new data source requires its own custom implementation, making truly connected systems difficult to scale.

MCP addresses this challenge. It provides a universal, open standard for connecting AI systems with data sources, replacing fragmented integrations with a single protocol. The result is a simpler, more reliable way to give AI systems access to the data they need.”*

At the highest level, MCP enables developers to build two-way connections between tools, services and data sources. The protocol consists of two parts, MCP servers and MCP clients.

MCP Server - The provider, which exposes specific functionality, like access to tools, services or data. The server knows how to receive a request and translate it to ensure compatibility with downstream systems.
MCP Client - The consumer, which runs on the application side, makes requests for tools, services or data. The client handles sending requests, receiving results, and passing them to the AI.

The MCP protocol manages this end-to-end process, ensuring everything is standardised and synchronised. It defines how the client and server communicate, what the messages look like, how actions are described, and how results are returned.

On the surface, this approach sounds like an Application Programming Interface (API), as an API and MCP both define how systems interact. However, an API is a general-purpose interface for various applications, whilst MCP is specifically designed to standardise how AI models, specifically Large Language Models (LLMs), interact with external tools, services and data.

Therefore, MCP does not replace the need for APIs. With this in mind, I expect we will see developers “wrapping” existing APIs for MCP.

Assuming widespread adoption, this could be a big step forward in how AI agents operate, unlocking the potential of Agentic AI.

Google Willow

Mon, 09 Dec 2024 00:00:00 -0600

I first started researching Quantum Computing in 2015, supported by an engagement with IBM to understand the potential for life sciences, specifically drug discovery.

Over the years, I have written a few articles summarising my key learnings.

As a brief reminder, in Classical Computing, a basic unit of information is called a bit, which is a binary digit that can have only one of two states, commonly represented as either a 0 or 1.

Quantum Computers use Qubits, which can also be set to two states, however, a Qubit can be in both states simultaneously, known as a superposition.

As a result, with 3 qubits of data, a Quantum Computer can store all eight possible combinations of 0 and 1 simultaneously. Therefore, a 3 Qubit-based Quantum Computer can complete calculations eight times faster than a 3-bit Classical Computer.

Another key property is entanglement, which describes the close connection between multiple Qubits, allowing a Qubit to instantly react to the change of another Qubit, regardless of their relative proximity. As a result, by reading the state of one Qubit, the state of its partners can be deduced without any additional checks.

These special properties help Quantum Computers solve complex problems much faster than Classical Computers.

Unfortunately, Quantum Computing is prone to errors, because qubits are sensitive to external disturbances, impacting the ability to scale reliably.

Earlier today, Google announced Willow, their latest quantum chip.

Willow boasts two important breakthroughs:

Willow can reduce errors exponentially as quantum processors scale up using more qubits. This targets the key challenge in quantum error correction.
Willow performed a standard benchmark computation in under five minutes that would take one of today’s most powerful supercomputers 10 septillion years.

These breakthroughs are an important step towards making Quantum Computing a viable alternative to Classical Computing for specific tasks.

To help understand the impact of this announcement, I would recommend reading the reaction from Scott Aaronson (Acclaimed Theoretical Computer Scientist).

I only understand the basic concepts of Quantum computing, but continue to find the subject fascinating and am pleased the community continues to invest and engage!

Even Realities

Sun, 08 Dec 2024 00:00:00 -0600

In September, I posted an article regarding the Ray-Ban Meta smart glasses.

I described them as a turning point for head-mounted devices, thanks to their “less is more” approach, delivering useful technology that is socially acceptable to wear for daily usage.

In the article, I referenced the Even Realities G1 smart glasses that like the Ray-Ban Meta were designed as glasses first, smart glasses second.

The Even Realities website states:

“Glasses you’d love, even if they were just glasses”

I have now had access to the Even Realities G1 smart glasses for a few months and therefore this article will share my initial thoughts and findings.

The Even Realities G1 smart glasses are very well built, with a premium look and feel. This includes robust materials (magnesium alloy, titanium alloy and silicone) and an innovative screwless hinge. They are available in a few different styles and colour options, with single-vision or prescription lenses that offer ultraviolet and blue light protection. They can also be converted into sunglasses, via a simple “clip” accessory.

I have access to the Even Realities G1 “A” smart glasses. The technical specification can be found below.

Micro-LED Display (640x200, 20Hz, 1000nits)
2x Microphones
2x TouchBars
2x Antennae (Bluetooth 5.2)
Integrated Battery (60 mAh / 0.616 Wh)
Case Battery (2000 mAh / 7.4 Wh)

The most interesting part of the specification is the Holistic Adaptive Optical System, known as HAOS.

HAOS incorporates Micro-LED optical engines that transmit content onto a pair of waveguide lenses, which then project information approximately two meters ahead of you.

Information is transmitted via “Dual Sync” connectivity, which includes a pair of antennae located in each hinge (behind the ear), communicating via Bluetooth 5.2. Impressively, each hinge also includes a TouchBar, allowing for physical input.

It should be noted that, unlike the Ray-Ban Meta, the Even Realities G1 smart glasses do not include a speaker. All information is delivered via the 640x200 display as green text.

At release, the Even Realities G1 smart glasses include the following features, configured via the companion iOS/Android app.

Quick Note
Translate
Navigation
Teleprompt
Notification
Dashboard
Artificial Intelligence (AI) Assistant

These are all software features, which I would expect/hope will continue to evolve and improve.

The build quality of the Even Realities G1 smart glasses is excellent, this includes the accessories and packaging, which are all thoughtfully designed, including specific decisions that promote sustainability.

I do not wear prescription glasses, therefore my experience is limited. However, the frames feel robust and I believe it would be very difficult for anyone to identify them as smart glasses.

The only noticeable design consideration is the bludge at the end of each hinge, which when worn sits behind each ear. Thankfully, the size and placement are discreet and could easily be considered a functional design choice to improve stability when worn.

Considering this is a v1.0 product, I have been incredibly impressed by the quality of the technology integration. This includes the iOS app, which is easy to navigate, fast and reliable.

Whilst wearing the smart glasses, the primary interface is a floating display, which can be activated at any time by looking up (this gesture can be adjusted in the app).

The default interface is the dashboard, which continues “at a glance” information, projected approximately two meters ahead of you. The interface remains visible for a pre-determined amount of time, depending on the specific feature being used.

The information does not obstruct your view (98% passthrough) and the position of the interface can be adjusted in the app. I would compare it to using a car heads-up display.

The TouchBars allow for simple inputs via single, double or triple taps, as well as a long press. Each input will trigger a different outcome, which is not immediately obvious, but fairly simple to understand once learnt. Full details of the TouchBar controls can be found within the Even Realities G1 User Guide.

The real magic occurs when a specific feature is triggered, for example, the ability to take a quick note, which includes integrated Generative AI summarisation.

I have been most impressed by the Teleprompt and Translate features, which work surprisingly well.

For example, with content loaded through the iOS/Android app, the Teleprompter will use the microphones to follow your spoken words, automatically progressing the teleprompt that is displayed ahead of you.

As someone who does a lot of live presentations and webinars, I could see how this feature could be valuable.

The same can be said for the translation feature, which listens for inputs and provides real-time subtitles via the display. This is very cool and a true glimpse into the potential of head-mounted devices.

An area that I expect to improve over time is the Artificial Intelligence (AI) Assistant. It would not be 2024, without an integrated AI feature. I have not found this to be as practical as a voice assistant (that speaks to you), but could be valuable for quick questions/answers.

In conclusion, I am incredibly impressed with the Even Realities G1 smart glasses, especially as this is a v1.0 product from a new company.

The design and build quality are impressive and the features they have enabled work as advertised, which is fairly surprising as many v1.0 products over-promise and under-deliver.

The Even Realiaities team should be commended for their attention to detail and I hope they continue to iterate/improve over time, unlocking the full potential of the smart glasses via new software features.

Reflecting on my Ray-Ban Meta usage, I would love to see the features of these two products combine, delivering audio, video, and display technology in a package that can be worn as a replacement for “regular” glasses.

This could be an incredible combination, that feels feasible, without impacting the design.

My Setup (Q4 2024)

Tue, 12 Nov 2024 00:00:00 -0600

Update - Hardware specification updated to include my new custom-built desktop PC and Framework Laptop 13 AMD AI 300 Series.

Final Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

I switch between three devices (one desktop and two laptops), which are primarily used for productivity, collaboration, software development, photo editing, video editing, virtual labs, gaming and game development.

My daily driver (home and travelling) is an Apple MacBook Pro 14-inch (NOV-2024). This laptop is also used to run Apple-specific software, such as Xcode and Final Cut Pro, as well as software that requires significant memory and/or memory bandwidth (e.g., local Large Language Models).

The full specification can be found below:

Apple MacBook Pro 14-inch (NOV-2024)
Apple M3 Max (16-core - 12 Performance / 4 Efficiency)
40-core GPU
16-core Neural Engine
128GB Unified Memory (546GB/s Memory Bandwidth)
2TB SSD (7.4GB/s Read)
3x Thunderbolt 5, HDMI 2.1, 3.5mm Headphone, SDXC Card Reader
14.2-inch Liquid Retina XDR Nano-texture Display (3024x1964 @ 120Hz)

In addition, I have a custom-built desktop PC. This is used to run software optimised for Windows, specifically games.

The full specification can be found below:

MSI MAG X870 Tomahawk WiFi
AMD Ryzen 9 9950X 4.3GHz Base / 5.7GHz Boost (16C/32T)
Noctua NH-D15 G2 LBC CPU Cooler
64GB Corsair Vengeance EXPO DDR5 PC5-48000C30 6000MHz RAM
2TB Crucial T705 M.2 PCI-e Gen5 NVM-e SSD
NVIDIA GeForce RTX 5090 Founders Edition 32GB GDDR7
Corsair HX1200i 1200W ‘80 Plus Platinum’ ATX PSU
Fractal Torrent Compact

Finally, I use an Framework Laptop 13 running Fedora for Linux-specific workloads (e.g., Linux Containers, Ethical Hacking, etc.)

The full specification can be found below:

Framework Laptop 13 DIY Edition
AMD Ryzen AI 9 HX 370 (12C/24T)
128GB Crucial DDR5-5600 RAM
4TB Western Digital Black SN850X NVMe (7.3GB/s Read)
13.5-inch LCD Display (2880x1920 @ 120Hz)

At home, the Apple MacBook Pro and desktop PC connect to a 57-inch Super Ultra-Wide monitor, specifically the Samsung G95NC Odyssey Neo G9 240Hz Dual UHD Monitor.

The full specification can be found below.

Model: Samsung G95NC Odyssey Neo G9
Panel Size: 57-inch
Panel Type: VA (Quantum Mini LED - 2,392 Zones)
Panel Curvature: 1000R
Aspect Ratio: 32:9
Resolution: 7680x2160
Refresh Rate: 240Hz
Variable Refresh Rate: AMD FreeSync Premium Pro
Response Time: 1ms (GTG)
High Dynamic Range: HDR10+ Gaming, VESA DisplayHDR 1000
Brightness: 420cd/㎡ (Typical), 1000cd/m2 (Peak)
Interface: 3x HDMI 2.1 (HDCP 2.2), 1x DisplayPort 2.1 (HDCP 2.2)

As you can see, for peripherals, I primarily use the NuPhy Air75 V2 (Wisteria Switches), Logitech MX Master 3S and Insta360 Link 2. The keyboard and mouse were selected as they support multiple Bluetooth profiles, making it simple to switch between devices. The mousepad is a Pulsar Superglide Glass Mousepad v2, which delivers very low friction and accurate tracking.

Additional peripherals include speakers (AudioEngine A1), headphones (Shure AONIC 5 Sound Isolating Earphones), controller (8BitDo Ultimate Controller), steering wheel (Logitech G29), virtual reality headset (Meta Quest 3 and Apple Vision Pro) and printer (Epson EcoTank ET-2810).

As highlighted in the photo below, my desktop PC is mounted to the desk using the Secret Lab Premium PC Mount, with the PC case door exposed providing easy access to the components.

Accompanying the desktop PC and monitor is an Original Prusa MINI+, which is an open-source 3D printer.

Finally, I use a Fractal Design Refine chair, which balances comfort and personalised ergonomics.

Home Server

Alongside my desktops and laptops, I have a small home server, which is connected to my Samsung S95B OLED TV (65-inch). The server manages local and cloud backups, server-side workloads, media streaming, and collaboration (video conferencing).

I selected an Apple Mac Mac Studio (MAR-2022) for the server, thanks to its excellent power efficiency (115W), high-performance I/O and small form factor. The Mac Studio also offers flexible video conferencing capabilities, including Apple FaceTime, Zoom, Microsoft Teams, Google Meet, Cisco WebEx, WhatsApp Video, etc.

Apple Mac Studio (MAR-2022)
Apple M3 Max (10-core - 8 Performance / 2 Efficiency)
32-core GPU
16-core Neural Engine
64GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
2TB SanDisk Extreme Portable SSD (USB 3.2 Gen2)

The 3TB of local SSD storage is split across multiple drives for resilience, with automated backups being completed by Carbon Copy Cloner and Cloud Storage via Microsoft OneDrive.

The only peripheral connected directly to the Mac Studio is a Logitech Brio Ultra HD Pro webcam, which delivers phenomenal video quality (4K/30fps - HDR), a wide viewing angle (90-degree dFoV) and a surprisingly good stereo, dual omnidirectional integrated microphone.

To support my minimal (hidden-wire) entertainment setup, all of my AV equipment (Sony STR-DN1050 AV Receiver, Sony PlayStation 5 Pro, Logitech Harmony Hub) is located in a cupboard under the stairs, with the required cables fed through the wall to the TV. This can be a little inconvenient when looking to insert/change physical media, but thankfully the cupboard offers plenty of space and ventilation.

Work Setup

Fri, 08 Nov 2024 00:00:00 -0600

In the article “My Setup (Q3 2025)” I outlined the systems I use at home, covering both personal and work use cases.

At work, I build upon my home setup, with the addition of two devices (one laptop and a desktop), specifically a Microsoft Surface Laptop 7 and a Framework Desktop.

The intent is to provide maximum flexibility for a wide range of workloads. For example, access to the three primary desktop operating systems (macOS, Windows, and Linux), as well as diverse bare metal hardware architectures (x86-64 and ARM).

The specification of the Microsoft Surface Laptop 7 can be found below:

Microsoft Surface Laptop 7
Qualcomm Snapdragon X Elite (X1E-80-100 - 12C)
32GB LPDDR5x (8448 MT/s)
1TB M.2 PCI-e 4.0 NVM-e SSD
Wi-Fi 7 and Bluetooth 5.4
13.8-inch LCD Touchscreen (2304x1536 @ 120Hz)
2x USB-C (USB 4), 1x USB-A (USB 3.1), 1x 3.5mm Headphone, 1x Surface Connect

This device, alongside my Apple MacBook Pro and HP ZBook Ultra G1a, was selected for its balance of performance, portability, and battery life. It also has an excellent high resolution, high refresh rate display, and a quality camera and microphone setup.

These capabilities are valuable, as I spend a lot of time jumping between different locations and working on virtual conference calls. This combination of hardware allows me to continue with almost any workload or testing regardless of my location, maximising my efficiency and effectiveness.

The Framework Desktop runs Fedora for Linux-specific workloads, including Linux Containers, Artificial Intelligence (AI), Ethical Hacking, etc. It also has a high-performance Accelerated Processing Unit (APU), with 128GB of unified memory, which is useful when running local AI models. The specification can be found below:

Framework Desktop
AMD Ryzen AI Max+ 395 3GHz Base / 5.1GHz Boost (16C/32T)
128GB LPDDR5x-8000 Unified Memory
AMD Radeon 8060S Graphics (40 Graphics Cores)
4TB Western Digital SN850X NVM-e SSD
2x Thunderbolt 4, 1x USB-C, 1x USB-A, HDMI 2.1, 3.5mm Headphone
AMD RZ717 Wi-Fi 7 (Wi-Fi 7 and Bluetooth 5.4)
Realtek RTL8126 5Gbit Ethernet
1x HDMI 2.1, 2x DisplayPort 2.1, 2x USB-C, 2x USB-A, 1x 3.5mm
2x Expansion Card Slots

When I am at my desk, all devices connect to a Samsung Odyssey OLED 49-inch Gaming Monitor (LS49DG912SUXXU), which has a 5120x1440 resolution at a maximum of 144Hz.

As you can see, for peripherals, I primarily use the Loitech MX Mechanical, Logitech MX Master 3S, and Opal C1.

I use two headsets, depending on the situation, the Bowers & Wilkins PX8 and Shokz OPENCOMM 2 UC

I am very fortunate to work in a very modern, open-plan office, which promotes collaboration.

Alongside being a great space, we have access to large meeting rooms, a cafe, a gym, and even a dedicated innovation room, which includes a range of hardware and software to promote curiosity.

DeskPad

Sat, 12 Oct 2024 00:00:00 -0500

At home, I use a super ultrawide monitor, specifically the 57-inch Samsung G95NC.

In my opinion, the 32:9 aspect ratio delivers the best overall experience when interacting with a desktop operating system. The scale of the working area streamlines most productivity tasks, by avoiding the need to constantly switch windows, whilst allowing for the “focus window” to remain centred.

Unfortunately, there are a few considerations when working with a monitor of this size, most notably, when you need to share your full desktop via a screen-sharing application, such as Microsoft Teams, etc.

In this scenario, unless the recipients of the screen share are fortunate enough to have a super ultrawide monitor, the content can become unusable.

Thankfully, Bastian Andelefski has resolved this issue for macOS users, via his simple (and genius) tool, known as DeskPad.

Once installed, DeskPad creates a virtual second display, which behaves like any other physically connected display. Therefore, you can adjust the settings (e.g., resolution) via system preferences and all application windows adjust accordingly.

As a result, you can simply select this virtual display when sharing your full desktop via a screen-sharing application.

Google Elevate Every Experience

Fri, 11 Oct 2024 00:00:00 -0500

Last week, I joined the Google Cloud team as a panellist as part of a webinar titled “Elevate Every Experience: AI solutions for marketing, support, and beyond”.

It was a fun discussion, with engagement from Best Buy and Gen Digital. We discussed many common themes, including the opportunities and risks associated with Generative AI, as well as our approaches and experience scaling solutions.

The webinar is now available on demand.

Thanks to the Google team for the opportunity to share our insights.

Intel Lunar Lake

Tue, 08 Oct 2024 00:00:00 -0500

In my recent article “Chip Wars”, I shared my perspective regarding the recent turbulence across the PC silicon market.

With the release of Intel Lunar Lake, formally known as Intel Core Ultra (Series 2), we now have the full picture, covering Qualcomm, Intel and AMD.

Specifically, I have had access to a Microsoft Surface Laptop 7 (Qualcomm), Asus Zenbook S 14 (Intel), and Asus Zenbook S 16 (AMD). Although these laptops are not like-for-like, they do offer insights into how each of the new architectures perform.

All laptops were equipped with 32GB RAM and a 1TB SSD. The specifications of each system-on-a-chip can be found below.

Qualcomm Snapdragon X Elite - X1E-80-100 (12C/12T, 4.2GHz Max, 45 TOPS)
Intel Core Ultra 7 258V (8C/8T, 4.8GHz Max, 47 TOPS)
AMD Ryzen AI 9 HX 370 (12C/24T, 5.1GHz Max, 50 TOPS)

On paper, the AMD Ryzen AI 9 HX 370 is the most powerful system-on-a-chip, with the most cores, threads, boost clock frequency and the highest rated Neural Processing Unit (NPU).

It is evident from the design that Intel has prioritised power efficiency with Lunar Lake, reducing the number of cores and removing hyper-threading. The Intel Core Ultra 7 258V has a total of eight cores, but only four are focused on performance.

Outlined below are the results from my testing, covering common and specialist workloads. Knowing that power can have a direct impact on performance, I also tested each laptop whilst on battery.

The first series of tests were completed when connected to mains power.

The second series of tests were completed when battery-powered.

As an immediate observation, the tests highlight a weakness of the Qualcomm Snapdragon X Elite regarding software compatibility due to the immaturity of Windows on ARM (something that I hope will improve over time). This can be seen with the results for PCMARK 10 (not compatible) and Tomb Raider (x86-64 translation).

Overall, I am excited to see that Qualcomm, Intel and AMD have all produced compelling products, each delivering significant improvements over the previous generation, with ~20% better performance, ~40% better battery life and dramatically lower thermal characteristics (reduced cooling requirements and fan noise).

In addition, they all meet the minimum requirement for “Copilot+ PC”, including a Neural Processing Unit (NPU) with a minimum of 40 TOPS. Currently, the real-world value of “Copilot+ PC” is minimal. However, I expect to see increased utilisation over the coming months/years. For example, there are rumours that Microsoft plans to offload Microsoft Teams features to the NPU, which should have a positive impact on performance and battery life.

Based on the results, it is evident that Qualcomm delivers exceptional processor performance (single and multi-threaded) when operating on mains power and battery. This is especially impressive, acknowledging the class-leading ~15 hours of battery life.

Therefore, if you are an individual who spends a lot of time working remotely on a laptop (no mains power or monitor connected), with a focus on productivity and collaboration tasks (e.g., Microsoft Office 365), Qualcomm offers a compelling value proposition. However, if you require specialist software or hardware, you may encounter compatibility issues. This creates a risk that should be carefully considered before any purchase (personal or business).

As highlighted in my previous article, x86-64 is still the best option for most people, delivering solid performance and battery life, with guaranteed software and hardware compatibility. Although Intel has taken a serious step forward with Lunar Lake, specifically regarding battery life and graphics performance, I still believe that AMD delivers the best overall value proposition.

Although AMD falls short on battery life (~10 hours), it excels in all other areas. Therefore, if you are a hybrid user, switching between docked (home/office) and remote working, with a requirement for broad software compatibility, I would recommend AMD.

The one challenge for AMD will be availability and manufacturer support. Historically, Intel has dominated in this area, with a limited number of AMD laptops. I suspect this is why Intel has chosen to target Qualcomm as the primary competitor with Lunar Lake, assuming that they can continue to outsell AMD (regardless of the laptop performance) based on availability, alongside the sheer number and variety of Intel laptops.

Google Gemini at Work

Tue, 24 Sep 2024 00:00:00 -0500

Over the past year, I have been documenting my team’s journey in developing and scaling a multi-model, multi-modal Generative AI framework for business, which has been used to power a wide range of use cases, primarily accessed via a natural language chatbot.

Earlier this week, Google Cloud CEO Thomas Kurian shared a few words about this initiative during the “Gemini at Work” global digital event opening keynote.

The keynote is available for on-demand viewing, with a brief spotlight on my team at around 27 mins. He references some key use cases, specifically pharmacovigilance, customer orders, and clinical insights, acknowledging the return on investment, which has now achieved an estimated $1.9M.

The Google Cloud YouTube channel (where the video is posted) has approximately 226,000 subscribers.

Finally, Thomas wrote an accompanying blog post, “Customers are putting Gemini to work”, in which he provided some additional detail regarding our work.

“Elanco, a world leader in animal health, has implemented a gen AI framework, powered by Vertex AI and Gemini, to support critical business processes, such as Pharmacovigilance, Customer Orders, and Clinical Insights. This has resulted in an estimated ROI of $1.9 million since launching last year.”

These references are great recognition for the team and hopefully help highlight the exceptional work across key domains, such as Science, Technology, Engineering and Maths.

Google AI Webinar

Fri, 20 Sep 2024 00:00:00 -0500

Update - The webinar is now available for on demand viewing.

On the 3rd of October at 18:00 UK, I will join the Google Cloud team as a panellist as part of a webinar titled “Elevate Every Experience: AI solutions for marketing, support, and beyond”.

The webinar will cover the use of Generative AI within business, including:

Create personalised campaigns, generate compelling content, and drive conversions with AI-powered tools.
Enhance productivity, improve self-service options, and deliver faster, more effective support.
Automate routine tasks, gain valuable insights from data and streamline workflows across your organisation.

The event can be watched live and will be made available for on-demand viewing.

Animal Health Innovation

Wed, 11 Sep 2024 00:00:00 -0500

My company is a global leader in animal health dedicated to innovating and delivering products and services to prevent and treat disease in farm animals and pets.

As the global Chief Technology Office (CTO) and Chief Information Security Officer (CISO), I keep a close eye on the wider industry, especially companies and/or community projects that are pushing the boundaries to unlock new value. This includes companies generating interesting data that could be used to identify or accelerate the discovery of new products.

Outlined below are some of the interesting companies/projects I am monitoring focused on animal health.

Companion Animal (Pet)

Butternut Box - D2C Pet Wellness
MI:RNA Diagnostics - Veterinary Diagnostics
Embark - Dog DNA Testing
AnimalBiome - Dog and Cat Gut Health for Digestive Disorders
Petpace - Pet Monitoring
Great Pet Care - Pet Management with Electronic Health Records
DVM Solutions - Zero-touch Vital Signs Remote Monitor
ALGENEX - Biotech Protein Production
Companion- Autonomous Dog Engagement via Computer Vision
Animab - Oral Monoclonal Antibodies
Exubrion Therapeutics - Chronic Pain in Companion Animals
Gallant Therapeutics - Allogeneic Cell Therapies for Chronic Conditions
ImpriMed - Personalised Drug Response Prediction Service for Cancer
MBF Therapeutics - DNA Vaccine Platform
LAIKA - AI for Veterinary Diagnostic Support
FitBark - Pet GPS and Health Tracker
Dog Aging Project - Canine Health Study
Benno Therapeutics - Cancer Treatment
Tractive - GPS and Health Tracker

Farm Animal

In Ovo - Poultry Animal Welfare
Biotangents - Livestock Infectious Diseases
Aviasenze - Poultry Sensor Technologies
Innovent - Pig and Cattle Sensor Technologies
SwineTech - Pig Sensor Technologies
Biomedit - Microbiome Innovation for Animal Health
Advanced Animal Diagnostics - Diagnostics to for Farm Animal Diseases
Performance Livestock Analytics - Livestock Management Solutions
Cainthus - Dairy Farm Technologies
OnCup AI - Livestock Management via Computer Vision
Serket - Livestock Health Management
Breedr - Livestock Supply-chain
Arbiom - Protein Source for Feed Manufacturers
Baseimmune - Universal Vaccines
Chordata Limited - Livestock Integrated Health Microchip
Cody Genetics - Genetic Editing Tools
Pondus - Poultry Monitoring and Measurement
VEROVACCINES - Poultry, Swine and Cattle Yeast-based Technology Vaccines
SellMyLivestock - Livestock Marketplace

Next year, I will attend the Animal Health, Nutrition, and Technology Innovation Europe event in London, hosted by Kisaco Research. This event provides an invaluable networking opportunity for the animal health industry, alongside specific channels for start-ups to showcase their ideas, products and services.

I am certainly looking forward to engaging with the community!

Chip Wars

Sun, 08 Sep 2024 00:00:00 -0500

Following years of stagnation, the CPU market has finally been revived.

It started with the launch of Apple Silicon for Mac in 2020, highlighting the potential of ARM for general-purpose computing. The performance, power management and thermal efficiency delivered by the Apple M1 System on a Chip (SoC) destroyed the competition, highlighting what was possible if ARM was taken seriously outside the mobile market.

Earlier this year, Qualcomm attempted the same manoeuvre targeting the PC laptop market (Windows and Linux), delivering the ARM-based Snapdragon X Elite/Plus SoC.

Although not quite as impactful as Apple Silicon, Qualcomm achieved their goal, demonstrating that it can disrupt the established duopoly of Intel/AMD.

Unsurprisingly, the success of this release, alongside strong, visible support from Microsoft and the hardware manufacturers, has forced Intel and AMD to rethink their strategies and increase their rate of innovation.

This led to the launch of AMD Strix Point in July and Intel Lunar Lake, announced last week.

I have tested AMD Strix Point and can confirm that AMD has successfully delivered performance, power management and thermal efficiency that rivals Qualcomm. Looking at the details shared by Intel, I expect them to achieve similar results with Lunar Lake (something I plan to verify personally).

Therefore, we now have three viable players delivering silicon for the PC laptop market, attempting to “one up” each other to solidify their position as the market leaders.

This level of aggressive competition is fun to watch, but more importantly great for consumers and businesses, helping to deliver notable innovation alongside competitive pricing.

With that said, the current market turbulence has also created confusion. Therefore, this article shares my perspective on the things to consider when purchasing a new PC laptop, covering x86-64 vs. ARM and Intel vs. AMD.

x86-64 (Intel and AMD)

100% compatible with existing software.
100% compatible with existing hardware, including printers, capture cards, eGPU docks, etc.
Support for discrete graphics cards from NVIDIA, AMD and Intel.
Excellent gaming compatibility and performance.

ARM (Qualcomm)

Compatible with mainstream software, covering web browsing, productivity, collaboration, etc.
Excellent thermal efficiency, resulting in low temperatures and fan noise.
Excellent idle efficiency, providing long standby times.
Excellent webcam quality.

In my opinion, the primary differentiator between x86-64 and ARM is software compatibility and support for discrete graphics cards.

Therefore, if you are a gamer or require specialist applications for your workloads (e.g., graphics, video), I recommend x86-64 (Intel and AMD).

If your workloads are primarily focused on web browsing, productivity and collaboration, leveraging mainstream applications (e.g., Office 365), I recommend exploring ARM (Qualcomm). This is especially true if you are a frequent traveller, recognising the thermal efficiency and long standby times offered by the ARM architecture.

Although I am pleased to see ARM finally become relevant for the PC laptop market, it is fair to state that x86-64 is still the best option for most people, helping to avoid any confusion regarding software/hardware compatibility.

Hopefully, over the next year, this distinction will become less relevant, with native ARM software and hardware drivers being made available alongside x86-64. In addition, the operating system (Windows or Linux) should be smart enough to automatically select the correct (native) architecture, without any user intervention (something that is not consistently managed today).

Assuming the intent is to purchase an x86-64 PC laptop, the decision is between Intel and AMD. Knowing that Intel Lunar Lake is not yet available, it is difficult to speak with confidence regarding the best option.

However, I would certainly not recommend purchasing an Intel-based PC laptop that pre-dates Lunar Lake, recognising that Intel is in the process of making a significant architecture transition.

I can speak with confidence that AMD Strix Point PC laptops, based on the Zen 5 architecture, are excellent. In my testing, AMD Strix Point delivers solid performance, power management and thermal efficiency.

AMD is commonly considered the underdog and does not have the same market presence as Intel. However, I would argue they are ahead regarding the quality and consistency of their architecture (e.g., Zen series). Unfortunately, AMD has been known to have issues regarding firmware and driver support, which can undermine their architecture advantage, especially within businesses, that require a solid firmware foundation for management, security, compliance, etc.

This is where Intel and their historical dominance of the PC laptop market has a clear advantage, supported by its tight working relationship with Microsoft and the hardware manufacturers. With that said, there is evidence that these critical relationships are not as robust as they once were, opening the door to greater competition from AMD and Qualcomm.

Therefore, as it stands today, I would happily recommend AMD Strix Point for general consumers, which will also have great Linux support as the relevant drivers get mainstreamed into the kernel. When Intel Lunar Lake becomes available, I expect it to rival AMD Strix Point but is unlikely to offer significant advantages. However, the wider support from hardware manufacturers will likely make Lunar Lake the preference for businesses.

Ray-Ban Meta

Sun, 01 Sep 2024 00:00:00 -0500

I have always been interested in wearable technology, specifically head-mounted devices.

Over the years, I have tested a wide variety of solutions, covering professional hardware such as RealWare, as well as consumer products, like Google Glass and Apple Vision Pro.

In my opinion, these devices have the potential to change the way we interact with technology via a concept known as ubiquitous/pervasive computing, which is where technology is an integrated part of common objects and is made to appear seamlessly anytime and anywhere.

If implemented effectively, this concept can make technology more personal, delivering value only when it is required, without becoming a distraction (a common issue with smartphones). For example, imagine walking up to a sign in a foreign language and having it automatically translated.

Unfortunately, every attempt to deliver a viable head-mounted device has failed in one key area, social acceptance.

For example, the term “glasshole” is used to describe anyone brave enough to wear Google Glass in public. Ironically, Google Glass was arguably the least offensive head-mounted device, when compared to products such as the Apple Vision Pro, etc.

Recently, I have been testing the Ray-Ban Meta smart glasses, specifically the Wayfarer model. What makes this head-mounted device interesting is that it looks and feels just like an ordinary pair of sunglasses.

In fact, I recently wore them to a family day out and not a single person noticed anything different, unique or special.

The embedded technology is fairly simple, specifically:

Camera (12MP)
Microphone
Speakers
Touchpad
Capture Button
Bluetooth and Wi-Fi

The Ray-Ban Meta smart glasses link to the Meta View app (iOS and Android), providing access to review photos/videos, whilst also triggering the the voice assistant.

Currently, the voice assistant is fairly limited (reading notifications, messages, etc.) However, this is an area where I expect to see continuous innovation, as Meta looks to integrate more Generative AI capabilities, leveraging the Meta Llama 3.1 language model.

At this point, I would expect the Ray-Ban Meta smart glasses to provide access to rich conversational content, similar to what we see from ChatGPT and Google Gemini.

As you would expect from a Meta product, the cameras include tight integration with Meta social services such as Facebook, Instagram, WhatsApp, Facebook Messenger, etc.

I don’t use any social networks, therefore I had feared the camera would be rendered useless. Thankfully, it can be used independently and privately, including offline (no Internet connection required).

The short video below shows my daughter and I going down a slide together. It was fun to be able to capture this moment, whilst also remaining completely present to enjoy the ride (not having to hold a smartphone, etc.)

The battery life of the Ray-Ban Meta smart glasses is rated at 4 hours from a single charge and up to 36 hours when combined with the included charging case. In my experience, I have been able to use the sunglasses on day trips without any issues, admittedly this was with limited/occasional usage of the smart features (notifications, messages, camera, etc.)

Recognising that the sunglasses are built by Ray-Ban, they are also very comfortable and excellent at protecting your eyes from the sun. They are even available with prescription lenses, meaning they deliver value, even when the technology is disabled or off. This is not the case with most wearable technology, which becomes useless when not active.

Finally, I think it is worth mentioning the newly announced Even Realities G1 smart glasses, which similar to the Ray-Ban Meta smart glasses, were designed as glasses first, smart glasses second.

Instead of delivering value via microphone and speakers, they include a system called HAOS (Holistic Adaptive Optical System). In short, Micro-LED optical engines transmit content onto a pair of waveguide lenses, which then project information approximately two meters ahead of you. This can include notifications, messages, navigation, and even real-time translations (e.g., subtitles).

I am excited to test the Even Realities G1 smart glasses but can see a world where these complimentary audio/video/display capabilities are combined. This could be an incredible combination, that feels feasible, without impacting the design.

In my opinion, this is a turning point for head-mounted devices, which has been achieved by acknowledging that “less is more”. For example, in most scenarios, users do not need the “ultra high tech” capabilities of an Apple Vision Pro. Instead, specific capabilities that directly target and enrich real-world scenarios are more valuable, whilst also making them viable for daily use and critically, and socially acceptable.

Product Operating Model Roles

Sat, 24 Aug 2024 00:00:00 -0500

In my previous two articles, I outlined the the Product Operating Model including details regarding the twenty product-first principles.

As a reminder, the Product Operating Model is a conceptual model, it is not a methodology, process or framework. It is about moving from output to achieving outcomes, following a set of product-first principles.

To enable The Product Operating Model, there are three roles that are critical when attempting to create a product that is valuable, viable, usable and feasible. These roles are commonly described as the product triad or product trio.

Outlined below are the key competencies and accountability for each role.

Product Manager

The product manager must have deep knowledge of the customer and the associated business processes. This includes strong situational awareness of opportunities, constraints, dependencies, etc. The product manager is accountable for ensuring any product is both valuable and viable.

Product Designer

The product designer (where required) participates in all phases of a product, from discovery to delivery to iteration. They focus on service design, user experience, prototyping, testing, and interactions. The product designer is accountable for ensuring the product is usable.

Technical Lead

The technical lead must have deep technical knowledge, including an understanding of principles, standards and positioning. The technical lead is accountable for ensuring the product is feasible.

These three roles are individual contributors, meaning they have no direct reports. As a result, there is one additional role (Product Leader) that supports the product triad/trio, focused on the people.

Product Leader

The product leader must have deep experience in staffing and developing individuals, requiring a holistic strategic context. The product leader is accountable for people management, team topology and communicating strategy, creating empowered individuals and teams.

It should be noted, that these roles are more than titles. The individuals associated with these roles must have the relevant experience and expertise to succeed. For example, taking a previous role, like a project manager, and renaming them to a product manager is likely to fail.

Product Operating Model Principles

Thu, 22 Aug 2024 00:00:00 -0500

In my last article, I outlined the Product Operating Model from the Silicon Valley Product Group (SVPG).

In this article, I will share more details regarding the twenty product-first principles.

Product Team Principles

As stated by Marty Cagan, “The most fundamental of all product concepts is the notion of an empowered, cross-functional product team.”.

Empowered with Problems to Solve: Product teams must be given a problem to solve, rather than solutions to build, and, most importantly, are held accountable for the results. They are empowered to make decisions and are not “order-takers”.
Outcomes over Output: Establish outcome-based goals, leveraging Objectives and Key Results (OKRs), which create focus on the priority, promote curiosity and provide relevant context. Outcomes are the results of actions, not features.
Sense of Ownership: Product teams must have the sponsorship and autonomy to come up with the solutions to the problems (Product Discovery), alongside the responsibility to build them (Product Delivery).
Collaboration: Promote respectful discourse, assuming good intentions when challenging ideas or decisions. Encourage a diverse perspective, but when a decision is concluded, ensure total commitment.

Product Strategy Principles

As stated by Marty Cagan, “The product vision is the future you are trying to create and, most importantly, how this vision will improve the lives of your customers.”

Focus: Ensure product teams have the foundations to focus, avoiding distractions that do not support the defined outcome. Be comfortable saying “no”, to promote discipline.
Powered by Insights: Leverage data, feedback and signals to inform strategy, ensuring strong situation awareness. Strategy is discovered, not pre-defined, requiring iteration and continuous improvement.
Transparency: Always start with why, ensuring the relevant context is communicated effectively to key stakeholders, building trust and credibility when making decisions.
Placing Bets: Acknowledge that there may be more than one solution to a given problem, consider multiple product teams with a rapid feedback loop to inform strategy and fail fast where needed.

Product Discovery Principles

Innovation occurs within Product Discovery and is a critical dimension on the road towards a valuable outcome. It is also the dimension that is commonly neglected in the race to start delivering.

Minimise Waste: Accept that failure can be a successful outcome of discovery, acknowledging that the majority (>75%) of ideas will not perform as desired. Attempt to solve every problem with the minimum amount of wasted time and effort.
Assess Product Risks: Every product team should have a common understanding of the risks, covering value, viability, usability, and feasibility.
Embrace Rapid Experimentation: Promote curiosity and empower experimentation with the product team, whilst ensuring the outcome is well-defined and remains relevant. Remove bureaucracy to avoid unnecessary friction, allowing teams to fail fast.
Test Ideas Responsibly: Experiments must be handled in a way that protects the company and its customers, covering concepts such as security, privacy, ethics, compliance, etc.

Product Delivery Principles

To promote empowerment and innovation, a single, cross-functional product team must be responsible for both discovery and delivery (not separate teams), ensuring continuity.

Small, Frequent, Uncoupled Releases: Leveraging an agile methodology, delivery should consist of short iterative sprints, with each sprint prioritising the important areas to deliver a viable product, always targeting the defined outcome.
Instrumentation: Establish the required technical foundations to observe how the product is used, providing valuable feedback to support continuous improvement.
Monitoring: Embed real-time monitoring at multiple levels to understand the product’s health, prompting short feedback loops that enable the team to proactively identify and rapidly respond to any concerns.
Deployment Infrastructure: Automate as much of the delivery lifecycle as possible, leveraging a standardised CI/CD pipeline to reduce human errors and increase team productivity.

Product Culture Principles

Excessive control is a primary risk factor preventing companies from innovating, growing and excelling, eroding trust and personal value.

Principles over Process: Process is great when both the problem and solution are known. However, the process can impact agility, making teams less responsive to new opportunities and threats, as well as more risk-adverse. These are all barriers to creating great products.
Trust over Control: As promoted by Simon Sinek, product teams can only be empowered when they are trusted, which requires psychological safety, supporting a “speak up” culture without any fear of humiliation or retribution.
Innovation over Predictability: 100% predictability = 0% innovation. Predictability is important only if it delivers the value the company depends on, targeting areas that are highly industrialised or commodity.
Learning over Failure: Failure only occurs when teams neglect the opportunity to learn or fail to apply the learnings to support future initiatives.

Additional detail regarding each product-first principle is outlined in the book TRANSFORMED, by Marty Cagan. I would also recommend The Product Compass, which is an incredible resource for Product Managers.

In the next article, I plan to share my thoughts regarding the key roles and responsibilities required to enable an empowered product team.

Product Operating Model

Tue, 20 Aug 2024 00:00:00 -0500

In 2020, we started our journey of implementing a product-led operating model, to increase our effectiveness and unlock new opportunities associated with digital/data business models.

The journey has not been straight forward and although we have made great progress, we continue to learn and evolve.

Earlier this year, Marty Cagan published his latest book called TRANSFORMED, which builds on the foundation for a Product Operating Model outlined in the first two books INSPIRED and EMPOWERED.

We have been leveraging this material, alongside direct engagement with Marty Cagan as our reference as we continue to mature.

Marty recently presented to our team, helping to grow our understanding further.

With this in mind, this article will outline the basics of the Product Operating Model, in the spirit of sharing our learnings with others who may be attempting a similar transformation.

The Product Operating Model from the Silicon Valley Product Group (SVPG) is a conceptual model, it is not a methodology, process or framework.

It is about moving from output to achieving outcomes, following a set of product-first principles.

Although individual company culture and outcomes can be very different, the product-first principles can be commonly applied or adapted.

The historical focus on output is a consistent root cause as to why companies fail to deliver value.

The Product Operating Model focuses on three dimensions:

Product Strategy: How do you decide what to work on? What is most important?
Product Discovery: How do you solve the problem? How do you discover a solution?
Product Delivery: How do you build, test and scale the solution? How do you prove the outcome?

Innovation occurs within Product Discovery and is a critical dimension on the road towards a valuable outcome. It is also the dimension that is commonly neglected in the race to start delivering.

There are twenty principles that support the three dimensions, including team and culture. They focus on ensuring any product is valuable, viable, usable, and feasible.

Product Team Principles

Empowered with Problems to Solve
Outcomes over Output
Sense of Ownership
Collaboration

Product Strategy Principles

Focus
Powered by Insights
Transparency
Placing Bets

Product Discovery Principles

Minimise Waste
Assess Product Risks
Embrace Rapid Experimentation
Test Ideas Responsibly

Product Delivery Principles

Small, Frequent, Uncoupled Releases
Instrumentation
Monitoring
Deployment Infrastructure

Product Culture Principles

Principles over Process
Trust over Control
Innovation over Predictability
Learning over Failure

Details of each principle are outlined in the book TRANSFORMED, by Marty Cagan.

In a future article, I plan to provide greater detail into the specific principles, including my own observations implementing at scale within a life sciences company with a seventy-year heritage that is not digital-first.

AMD Ryzen AI 370 HX

Mon, 19 Aug 2024 00:00:00 -0500

This is a very interesting time for Windows-based laptops, with ARM-based architecture becoming viable, new processors from Qualcomm (Snapdragon), AMD (Strix Point) and eventually Intel (Lunar Lake), alongside new hardware/software capabilities focused on Artificial Intelligence (AI).

Over the past few months, I have been testing the new Qualcomm Snapdragon X Elite, which is ARM-based, not x86-64. Specifically, I have access to the Microsoft Surface Laptop 7 and Surface Pro 11. Overall, I have been very impressed by the results, with both devices delivering excellent performance, battery life and thermal efficiency in a compact design (equivalent to the Apple MacBook Air, etc.)

As a direct response, AMD recently released the AMD Ryzen AI 300 series of processes (codename Strix Point), which are the first to be based on their next-generation Zen 5 architecture. In theory, these new processors will deliver for x86-64 laptops all of the benefits offered by the ARM-based Qualcomm Snapdragon X Elite.

I have been testing the AMD Ryzen AI 9 HX 370 (horrible name), which is the flagship mobile processor, targeting a configurable TDP (cTDP) of 15-54W.

The AMD Ryzen AI 9 HX 370 is built on the TSMC 4nm FinFET technology and includes 4x Zen 5 and 8x Zen 5c cores, providing a total of 12 physical cores and 24 threads. The processor can deliver a single-core boost speed of up to 5.1GHz for the Zen 5 cores and up to 3.3GHz for the Zen 5c cores.

This is the first time AMD has produced a “big.LITTLE” design, with the inclusion of Zen 5c cores, which should help deliver better battery life and thermal efficiency across common workloads (e.g., productivity, collaboration, etc.)

In addition, the AMD Ryzen AI 9 HX 370 includes the new AMD Radeon 890M graphics, built on the RDNA 3.5 architecture, with 16 cores and a frequency of 2.9GHz.

Finally, the Neural Processing Unit (NPU) has been improved, delivering 50 TOPS for local AI workloads. This is above the Microsoft Copilot+ PC requirement of 40 TOPS, but AMD devices are not yet Copilot+ PC branded.

Overall, this specification is very impressive and certainly (on paper) rivals anything offered by Qualcomm or Intel. Full details regarding the specification of the AMD Ryzen AI 9 HX 370, including connectivity options, can be found on the AMD website.

I have been testing the AMD Ryzen AI 9 HX 370 with the Asus ProArt P16. The full specification can be found below:

Asus ProArt P16 (H7606)
AMD Ryzen AI 9 HX 370 (12C/24T)
64GB LPDDR5x (7500 MT/s) Memory
2TB NVMe M.2 SSD
NVIDIA GeForce RTX 4070 8GB GDDR6 (105W)
16-inch WQUXGA OLED Touch Display (3840x2400 @ 60Hz)
Wi-Fi 7, Bluetooth 5.4

The Asus ProArt P16 is a high-performance laptop designed for creative workloads (hence the ProArt branding). However, it is surprisingly compact considering the specification (including a discrete GPU) and the 16-inch OLED display.

It also includes a massive array of connectivity, including USB 4 Type-C, USB 3.2 Gen 2 Type-A, HDMI 2.1, SD-Express 7.0, etc. This makes the Asus ProArt P16 highly versatile, removing the need for any dongles or hubs.

The focus of this article is the AMD Ryzen AI 9 HX 370. However, I will state the built quality of the Asus ProArt P16 is excellent, supported by premium materials and a dense, rigid chassis. The keyboard and large trackpad are both very good (unfortunately, the trackpad is not haptic).

My only critique regarding the hardware is the 60Hz refresh rate, which is acceptable, but for a laptop of this quality, it would have been nice to see a higher refresh rate (e.g., 120Hz).

If you are interested in the Asus ProArt P16, I would recommend watching the review from Andres Vidoza.

Outlined below are the performance numbers for the AMD Ryzen AI 9 HX 370, running on Windows 11 24H2 (26100.1457) and the “standard” power profile.

The numbers are compared against the Microsoft Surface Laptop 7 and Microsoft Surface Pro 11, running the Qualcomm Snapdragon X Elite (X1E-80-100) and the Apple MacBook Air 15-inch, running the Apple M3.

The first series of benchmark results were running on mains power.

The second series of benchmark results were running on battery.

As you can see, the single-core and multi-core performance numbers in Geekbench 6 are good. They are comparable to the Qualcomm Snapdragon X Elite but fall slightly below the market-leading Apple M3 for single-core performance.

The performance when running on the battery was also competitive against the Qualcomm Snapdragon X Elite. This is a positive sign, as in my experience, older AMD and Intel (x86-64) processors include a significant performance impact when not connected to mains power.

Amazingly, the Apple MacBook Air continues to demonstrate its dominance regarding power efficiency, delivering nearly 100% of the performance regardless of the power source (which is remarkable).

Graphics performance is where the AMD Ryzen AI 9 HX 370 demonstrates its strength, with the integrated AMD Radeon 890M graphics delivering double the performance of the Qualcomm Snapdragon X Elite and easily outperforming the Apple M3.

The integrated AMD Radeon 890M graphics power their way through productivity and collaboration workflows without any hiccups, and can even play modern games (1080p) at respectable quality settings and frames per second. This is an exciting prospect for future ultra-portable and handheld gaming devices (like the Steam Deck) that could be based on the AMD Ryzen AI 300 series of processors.

In addition to the excellent integrated graphics performance, the AMD Radeon 890M can also output high resolution, and a high refresh rate via HDMI 2.1, including my Samsung G95NC Odyssey Neo G9. This was not possible with the Qualcomm Snapdragon X or Apple M3 (the Apple M3 Pro and M3 Max are both capable of 7680x2160 at 120Hz).

Regarding thermal efficiency, the Asus ProArt P16 includes a standard heatsink/fan setup. This is notable, as comparable Intel-based laptops from Asus are usually equipped with a more robust vapour chamber. Even with the heatsink/fan setup, the AMD Ryzen AI 9 HX 370 runs reasonably cool (on average at 76C), with minimal fan noise during common workloads (e.g., productivity and collaboration). The fans are audible during more demanding tasks, such as gaming, however, this is to be expected and (in my opinion) acceptable.

Finally, although I have not tested it objectively (I simply use the laptop) the battery life of the AMD Ryzen AI 9 HX 370 has been very good, rivalling the Qualcomm Snapdragon X Elite. This is a major step forward for x86-64 laptops.

With that said, my initial testing highlights that the Qualcomm Snapdragon X Elite is more efficient when on standby (< 5% battery drain overnight). By comparison, x86-64 laptops are notoriously bad at maintaining battery life when not in use (in your bag or overnight). This can be a huge pain when travelling and therefore still an important consideration for anyone who values long standby times.

The AMD Ryzen AI 9 HX 370 does appear to deliver better standby efficiency than previous generations but still drains at a higher rate than I would want.

In conclusion, I believe AMD has successfully produced a mobile processor that delivers excellent performance, specifically graphics performance, alongside strong battery life and thermal efficiency (low fan noise).

In my testing, the AMD Ryzen AI 9 HX 370 rivals the Qualcomm Snapdragon X Elite, which was the target, falling slightly below the Apple M3 as an overall package.

This is an excellent result, which combined with the border compatibility for x86-64 software in the Windows ecosystem and the option to include discrete graphics, such as the NVIDIA GeForce RTX 4070, makes the AMD Ryzen AI series the best overall processor for most people.

At this time, I have not fully tested the business compatibility (e.g., firmware stability, secure enclave). However, assuming everything works as advertised, the AMD Ryzen AI series feels like the best balance for performance, efficiency and software compatibility.

Over the coming weeks, I plan to continue my testing, including targeting Linux, where I believe the AMD Ryzen AI series could shine!

Generative AI - Update

Tue, 13 Aug 2024 00:00:00 -0500

Over the past year, my team developed and scaled a multi-model, multi-modal Generative AI framework for business, which has been used to power a wide range of use cases, primarily accessed via a natural language chatbot.

I have documented this journey via a series of articles (links below).

In my last article, I shared six diverse business use cases and their outcomes, which contribute towards the wider return on investment, primarily focused on productivity acceleration and data insights.

Earlier today, the team launched the next major revision of the Generative AI framework, including a wide range of new features, as well as a new user experience for the natural language interface.

Outlined below is a summary summary of the new features.

A refined and simplified user experience that includes user personalisation, smart recommendations (tooltips), and light/dark modes.
The ability to start and maintain multiple conversation threads, each targeting a specific persona, model and context.
New multimodal models that can now handle up to 2,000 pages of text or 100 minutes of video content. Text and video content can be linked or uploaded directly.
New interactive user interface elements, including code blocks, tables and text, which can be easily exported, automatically selecting the appropriate file format (e.g., CSV for tables, etc.)
Access to the latest and greatest foundational models from OpenAI and Google with no restrictions. This includes the Imagen v3 for advanced image generation.
New APIs for developers, allowing the Generative AI framework to be programmatically integrated with any system.
New native integrations, covering systems of record, such as SAP and Microsoft SharePoint.
New getting started information, providing education material, frequently asked questions and examples for common use cases.

The image below highlights the new user experience (dark mode), including the customisable chat pane, tooltips and improved discoverability of custom personas.

Our custom persona capability (similar to OpenAI GPTs) also received a major upgrade.

Custom personas allow any user to create and publish their own Generative AI chatbot (no technical skillset required), leveraging the underlying framework. These custom personas can be published to individuals, teams or the entire company, secured (where needed) by modern authentication/authorisation (FIDO2 compliant).

A custom person can target any available model (e.g., OpenAI, Google) and include specific grounding and context, including large datasets from structured or unstructured data sources.

In the latest release, custom persona creation has been dramatically simplified via a step-by-step wizard that can be previewed in real time.

This new approach helps ensure every custom persona is configured correctly with optimal inputs to improve the quality and consistency of the output. It also includes native integrations to securely and privately ingest large datasets from systems of record, such as Microsoft SharePoint.

Finally, where appropriate, the custom personas can be posted to the marketplace, promoting discoverability, collaboration and reuse, which further increases the return on investment.

At the time of writing, 773 active custom persons, covering everything from business-specific expert systems to general productivity accelerators for common tasks such as knowledge sharing, summarisation, translations, etc.

Overall, our Generative AI framework continues to gain momentum, with more than 6,700 unique users (95% of the target audience) and approximately 1,900 logins per day. Conservatively, since launch (one year ago), this has saved approximately 35,564 hours (1,482 days) and delivered an estimated return on investment of $1.7MM through productivity acceleration and cost avoidance.

The image below highlights the dynamic dashboard, showing the growth over the past year.

Although we have seen incredible viral growth over the past year, we continue to believe this is just the beginning and therefore, we are excited to see what comes next!

Stevie Awards 2024

Thu, 08 Aug 2024 00:00:00 -0500

Following the success of the recent Globee Awards 2024, I am pleased to share that the team also won a Bronze Stevie Award in the category of “Best Business Technology Pivot” at the 22nd Annual American Business Awards.

The award recognised our new omnichannel digital engagement strategy, grounded in data and artificial intelligence (AI), which successfully increased customer touch points by 30%, reducing the average time to first purchase by 28%.

The Stevie Awards honour and generate public recognition of the achievements and positive contributions of businesses and working professionals worldwide.

The winners were determined by the average scores of more than 300 professionals worldwide in the three-month judging process.

Congratulations to the team and all of the award winners!

Globee Awards 2024

Thu, 01 Aug 2024 00:00:00 -0500

Earlier today, my team were recognised at the 19th Annual 2024 Globee Awards for Technology!

The team won two awards, a “Gold” in the category of “Achievement in Digital Transformation” and a “Silver” in the category of “Achievement in Enhancing Business Operations”.

The “Digital Transformation” award recognised the work associated with Digital Vet Engagement, specifically a new omnichannel strategy, grounded in data and artificial intelligence (AI), that increased customer touch points by 30% and reduced the time for first purchase during a product launch by 28%.

The “Enhancing Business Operations” award highlighted the impact of our private, secure and compliant Generative AI chatbot, which accelerates productivity by streamlining common business tasks. Since its launch, the chatbot has received approximately 1900 daily logins and has delivered an estimated return on investment of $1.7MM through cost avoidance.

The term “Globee” is a combination of the words “global” and “business”. The Globee Awards were formed to honour organisations of all types and sizes from around the world for their achievements in various business and technology-related categories. The awards were created to provide a platform for recognising and promoting excellence in industries and sectors that are shaping the future of business.

The judging process is rigorous and comprehensive to ensure fairness and integrity in evaluating entries. After the submission deadline, a panel of experienced judges from diverse industries and backgrounds is carefully selected. These judges possess expertise and knowledge in their respective fields, ensuring a thorough assessment of the entries.

Congratulations to the team and all of the award winners!

Surface Pro 11

Tue, 23 Jul 2024 00:00:00 -0500

Last month, I posted my initial impressions of the Microsoft Surface Laptop 7, running the new Qualcomm Snapdragon X Elite processor.

This week, I have been testing the Microsoft Surface Pro 11, which is a 2-in-1 (hybrid) device, part of the Copilot+ PC range.

In this article, I will share some initial impressions and comparisons with the Surface Laptop 7.

The specification of the Surface Pro 11 can be found below:

Microsoft Surface Pro 11
Qualcomm Snapdragon X Elite (X1E-80-100 - 12C)
32GB LPDDR5x (8448 MT/s)
1TB M.2 PCI-e 4.0 NVM-e SSD
Wi-Fi 7 and Bluetooth 5.4
13-inch OLED Touchscreen (2880x1920 @ 120Hz)
2x USB-C (USB 4), 1x Surface Connect

This specification is comparable to the Surface Laptop 7 I used for testing in my previous article. Therefore, I would expect similar performance and power-efficiency characteristics.

The hardware and build quality of the Surface Pro 11 is excellent, consistent with the Surface Laptop 7.

The two primary hardware differences are the chassis and display.

The Surface Pro 11, being a a 2-in-1 (hybrid) device, can be used as a standalone tablet or a laptop replacement, by connecting a keyboard via the proprietary port. As a result, the chassis is smaller and thinner than the Surface Laptop 7 (just 9.3 mm thick).

The Surface Pro 11 is comparable to the Apple iPad Pro, but arguably more versatile, as the device runs a full desktop operating system (Windows on ARM) vs. a mobile-orientated operating system (iPadOS).

I have been using the new Microsoft Flex keyboard, which connects directly via the proprietary port or Bluetooth and includes a stylus, which can be neatly and securely stowed in the keyboard when not in use (far superior to the Apple magnetic option).

The keyboard itself is good and includes a haptic trackpad, which is relatively small but delivers a precise, consistent experience. My only (minor) critique is the flex that occurs when typing at speed, something that Apple recently resolved with the Magic Keyboard for iPad Pro, via the use of aluminium.

The display is a different story. Being OLED, I had high expectations and had assumed it would deliver a notable improvement over the LCD panel used by the Surface Laptop 7.

The colours and black levels are impressive (to be expected with OLED). Unfortunately, these benefits are undermined by what I assume to be a visible digitiser (touch display layer), which impacts the image quality, making everything appear “grainy”. It is especially noticeable on white/grey backgrounds.

Initially, I thought I had a defective device. However, Reddit posts confirm this is by design.

I suspect this characteristic will impact users differently. Some will simply learn to ignore it, whilst others will find it to be a constant distraction. Unfortunately, I do find it distracting and therefore find it difficult to recommend the Surface Pro 11.

I have not tested the non-OLED version of the Surface Pro 11, therefore cannot confirm if it also suffers from the same image quality issue.

Regarding performance, The Surface Pro 11 generally competes with the Surface Laptop 7 (both with the Qualcomm Snapdragon X Elite, 32GB RAM, 1TB Storage).

However, as highlighted by the numbers below, I noticed an anomaly when running the Surface Pro 11 on the “Recommended” preset, whilst connected to Power.

Instead of delivering the full performance, the Surface Pro 11 throttles to the equivalent performance on battery. The Surface Laptop 7 does not exhibit this same characteristic, always delivering full performance (regardless of the preset) whilst on power.

It is unclear to me if this is a bug or a design decision. For example, the thinner chassis of the Surface Pro 11 may not be as thermally efficient. Therefore, the Microsoft team may be purposely limiting the performance to manage temperatures.

With that said, manually switching to the “Best Performance” preset resolves the “issue”. Delivering the expected performance when compared against the Surface Laptop 7.

Overall, the Surface Pro 11 is an excellent device, likely the best Surface Pro ever. However, the display quality issue on the OLED version is in my opinion, a major disappointment. It is so significant, that I highly recommend trying the device before you buy to ensure you are comfortable with the trade-off.

CrowdStrike IT Incident

Sat, 20 Jul 2024 00:00:00 -0500

Update - Microsoft has released an official recovery tool to support impacted businesses and users.

Over the past 48 hours, we’ve witnessed one of the largest global IT incidents in recent history, caused by a software bug from the security company CrowdStrike.

Microsoft has predicted that 8.5 million Windows systems have been impacted, disrupting commerce, travel, healthcare, manufacturing, advertisement and many more industries.

Specifically, on the 19th of July, as part of ongoing operations, CrowdStrike released a sensor configuration update to Windows systems.

Sensor configuration updates are an ongoing part of the protection mechanisms of the CrowdStrike Falcon platform.

The configuration update triggered a logic error (likely a null pointer exception) resulting in a system crash and blue screen of death (BSOD) on impacted systems, which persists on reboot.

Thankfully, my company was not directly impacted. Therefore, I thought I would explain why.

As part of our enterprise architecture principles, we prioritise vertical integration for endpoint security.

This means, wherever possible, we leverage the native capabilities of the underlying platform (e.g., Windows, Linux, macOS, etc.)

This strategy has three primary benefits:

An improved user experience, avoiding the need for multiple third-party software packages that consume resources, impacting performance and reliability.
Simplified, lower-cost support, removing the need for IT to procure, manage and lifecycle multiple independent security software packages.
Improved security posture, acknowledging that third-party software focused on security commonly requires kernel-level operating system access. Ironically, this makes security software a prime target for cyber-attacks and a continuous software assurance risk.

It is this strategy that protected us from the recent global IT incident.

It would be easy to blame Microsoft for the state of Windows security and they should take partial accountability. However, due to a series of flawed regulatory and legal controls, society must also take responsibility, as it is these controls that have “forced” the use of third-party security software, arguably making the underlying operating system less secure/reliable.

Steven Sinofsky (ex-president of the Windows division at Microsoft) shared the following tweet (which I agree with).

With that said, our strategy is not perfect and like all things, has its trade-offs. Therefore, we must continue to make smart decisions to ensure these trade-offs are understood and/or mitigated via other technologies and/or techniques.

I empathise with the global IT community, who will likely be in the spotlight for days/weeks as they work diligently to recover from this incident.

Finally, this incident is a warning for the IT industry (and the world), highlighting the fragility of the interconnected ecosystem, specifically as it relates to common foundations, such as Microsoft Windows, etc.

If you are impacted by this incident, I recommend reviewing the official CrowdStrike documentation for details regarding the recovery procedure.

As a reminder, cybercriminals will likely attempt to use this incident as a channel of attack. Therefore, only reference and/or engage with official (verified) documentation from CrowdStrike or a trusted partner.

Collecting Dust

Wed, 03 Jul 2024 00:00:00 -0500

It has been approximately five months since the launch of the Apple Vision Pro.

In the conclusion of my initial impressions, I stated “The Apple Vision Pro is the most impressive VR/AR headset I have ever used! It is also the most expensive and uncomfortable, whilst currently offering very little real-world value.”

Apple marketing states that the Apple Vision Pro delivers the following value proposition:

Productivity - A workspace with infinite space
Entertainment - The ultimate cinema
Photos and Videos - Be in the moment
Connection - Share quality time
Apps - Do what you love

In reality, even after five months of ownership, I do not believe the Apple Vision Pro has come close to meeting these expectations, meaning it has quickly become an expensive dust collector.

In this article, I will attempt to explain my rationale for this position.

Productivity

In theory, the Apple Vision Pro has the potential to be a great productivity enabler, delivering exceptional passthrough video for augmented reality and (when needed) full 360-degree virtual environments that can be customised for any situation.

Unfortunately, the software is currently a limiting factor. The native apps are not yet well positioned for content creation and the Mac Virtual Display capability (macOS screen sharing), which you would assume would be well positioned to leverage the potential of an “infinite workspace”, is actually very finite. For example, even basic options such as the ability to set multiple displays or to select a portrait orientation are not available, making it far less versatile than the Virtual Desktop app, available for the Meta Quest.

Although the Mac Virtual Display capability is expected to improve with visonOS 2.0, the ongoing restrictions regarding display size (the equivalent of two 4K displays), format (orientation) and connectivity make it less practical than a traditional monitor, specifically a multi-monitor setup or a super ultra-wide.

As a comparison, at home, I use a Samsung 57-inch G95NC Odyssey Neo G9 super ultra-wide monitor, which has a resolution of 7680x2160 at 240Hz and can be connected to multiple devices (Mac, PC, Video Game Consoles, etc.) At approximately half the price of the Apple Vision Pro, I would argue this monitor delivers better real-world value, with enough savings to purchase a new Apple Mac or PC.

The Apple Vision Pro does have the advantage of being virtual (no physical space requirement), but this benefit is dramatically reduced by the fact that you have to wear a cumbersome and often uncomfortable headset, which for long productivity sessions (anything more than a couple of hours) is far from ideal.

The (unflattering) photo below shows my face after approximately 30 minutes of using the Apple Vision Pro with the Solo Knitted Band.

As you can see, there are visible pressure marks on my forehead, nose and cheeks. I find the nose pressure the most uncomfortable. The comfort is improved when using the Dual Loop Band, but not completely resolved.

Entertainment

The Apple Vision Pro is an impressive entertainment device. For example, the Apple TV cinema environment is incredibly immersive, especially when watching content designed for 3D.

However, when I watch a movie at the cinema or a TV series at home, it is usually with family or friends. In my opinion, part of the experience is enjoying the entertainment with others, reacting to their reactions, whilst also having the opportunity to discuss afterwards. With the Apple Vision Pro, unless you happen to have multiple headsets, the experience is very isolating.

Even when I do occasionally watch a movie alone, the effort associated with wearing the headset for multiple hours, whilst plugged into a power source (due to the battery limitations) is a major barrier. Realistically, I can not see many people accepting this trade-off unless they are highly motivated to experience very specific content made for the Apple Vision Pro.

In conclusion, similar to other VR/AR headsets I have used, such as the Meta Quest series, I have found short experiences (less than 30 minutes) to be the sweet spot for the Apple Vision Pro. Anything longer quickly becomes a burden.

Photos and Videos

Apple has been eager to highlight the potential of spatial photos and videos as part of their marketing material. I can confirm, the demo content provided by Apple is very impressive. The spatial video in particular can feel very immersive, best described as “peering into a dream”. The illusion is not perfect, but good enough to trick the mind, triggering an emotional response.

Unfortunately, I have been less impressed with the content I have captured, either directly from the Apple Vision Pro or via an Apple iPhone 15 Pro. My content has a tendency to feel “flat”, failing to deliver the same level of immersion as the demo content.

It is very possible that my camera skills are the limiting factor. However, I would have to assume that most people are not professional photographers and/or videographers. Therefore, I fear my experience may be closer to reality. It is also possible that the demo content was captured with more advanced hardware/software.

Although I can see scenarios where having spatial photos and videos would be interesting (family milestones, events, etc.) The fact that they can only be viewed whilst wearing the headset makes them very limiting and the inconsistent quality I have experienced makes me question if having a higher-resolution 2D version would offer greater long-term flexibility. I do expect this story to evolve/improve over time, with new capabilities to translate 2D photos/videos into spatial photos/videos via artificial intelligence.

Connection

Apple offers several options to interact with others whilst wearing the headset, both directly (physically face-to-face) and via virtual channels (FaceTime, etc.)

The first is “EyeSight”, which is the controversial feature that attempts to show a virtual representation of the wearer’s eyes on the outward-facing lenticular display. The intent is for the wearer to be able to naturally interact with people in close proximity, without the need to remove the headset. Although I commend Apple for its creative thinking and technical innovation, the practical usage of this feature is poor. I have yet to find a scenario where anyone interacting with me in the physical world whilst wearing the headset is not distracted, completely breaking any face-to-face connection.

The virtual experience has a similar issue. Apple has developed a way for the wearer to create a virtual “Persona”. The technology itself is incredible and the Persona can be used across multiple apps, including FaceTime, Teams, Zoom, etc. However, with the best will in the world, the Persona is not a perfect representation, resulting in an “uncanny valley” effect. Although the Persona attempts to replicate expressions, it simply lacks the fidelity to deliver the complexity and nuance of human emotion.

Therefore, similar to EyeSight, interacting with people in virtual meetings using my Persona, quickly becomes a distraction. I have had comments like “It is like I am looking at your ghost” and when interacting with family members, their feedback suggests that they would prefer a standard video feed, where they can see my actual face. In summary, the Persona does not enhance the connection and for that reason, I not yet had the confidence to use the Apple Vision Pro during any serious meeting (specifically where the participants are not wearing Apple Vision Pro).

In fairness to Apple, I do believe the Apple Vision Pro has the potential to enhance the connection between people, especially virtually. However, I fear the technology is not yet advanced enough to deliver an experience that most would consider acceptable for frequent use.

Apps

At this time, the apps available for the Apple Vision Pro are mostly limited to technology demos. These are usually short “experiences” that can deliver an instant “wow” factor but offer very little real-world use.

It is great to see native support for mainstream apps, such as Microsoft Word, PowerPoint, Excel, etc. However, these apps are limited versions of their desktop/mobile counterparts, making them great for viewing content, but less useful when attempting to create/modify. As a result, similar to the common complaint about Apple iPadOS, the software restrictions of Apple Vision Pro are limiting the total value proposition.

Recognising the high price-point and limited availability of the Apple Vision Pro, it is not a surprise that the developer adoption has been slow. However, I continue to be a little confused as to why Apple has not delivered more compelling first-party apps. I feel there is an opportunity to lead by example, helping to inspire the developer community and giving early adopters a reason to use the headset on a more regular basis.

Conclusion

Overall, I appreciate my comments after five months are not overly positive and I continue to stand by my initial observation that the Apple Vision Pro is the most impressive VR/AR headset I have ever used! However, it is also the most expensive and uncomfortable, whilst currently offering very little real-world value.

With that said, I am still incredibly optimistic and excited by the future of VR/AR and spatial computing. However, I fear we are many years away from it becoming a viable mainstream capability that can replace existing usage patterns, achieved today with smartphones, tablets, laptops, etc.

As a result, after the initial excitement of testing the Apple Vision Pro, the incentive to use it on a regular basis has quickly faded. I did attempt to work within the Apple Vision Pro for a full day, but the previously highlighted barriers were simply too great for the experience to be practical and it is honestly difficult to highlight any use case that was “enhanced” by wearing the headset.

Therefore, I continue to believe, that if you are interested in VR/AR today, I recommend exploring the options from Meta (specifically the Meta Quest 3), which is more comfortable, has a broader software library and is a fraction of the cost at just £479.99 vs. £3,499.

Finally, earlier this week, I was asked to join Apple for a private meeting at their Battersea Power Station office to discuss the launch of the Apple Vision Pro in the UK, specifically regarding opportunities within business.

I predict that most people who were invited for a private briefing had very little previous hands-on experience with the Apple Vision Pro. Therefore, I suspect my experience and feedback were very different to the standard. As previously stated, the first impression when using the Apple Vision Pro is incredible, delivering a real “wow” factor. However, it is only when used over time (a couple of hours) that you start to identify the issues. Therefore, I suspect most feedback was positive vs. the higher level of constructive critique that I provided based on my usage over five months.

I do not expect my feedback to have any meaningful impact, but I do hope that Apple continues to listen/learn from their passionate community regarding how to iterate/improve on the Apple Vision Pro, even if it requires them to pivot on previous decisions.

Specifically, I would recommend the following.

Remove the front-facing display.
Replace heavy premium materials with lighter materials, reducing the overall weight.
Consider function over form (design), recognising that a device you wear on your face must be comfortable.
Reduce the cost significantly (target £999), making hardware cutbacks where needed (e.g., front-facing display, premium materials, etc.)
Simplify the initial setup and offer multiple user accounts (not just guest accounts).
Deliver compelling first-party apps that lead by example, highlighting the potential of the hardware.
Incentivise the developer community to build interesting apps that offer real-world value (not technology demos), supported by better access to developer-specific hardware.

I am interested to see how Apple will proceed!

Surface Laptop 7

Mon, 24 Jun 2024 00:00:00 -0500

Update - In the article, I highlight a Microsoft Defender compliance issue. It would appear this issue is caused by the Windows Home to Pro upgrade process. The command “DISM /online /Add-Capability /CapabilityName:Microsoft.Windows.Sense.Client ~~~~” executed as an administrator, followed by a restart, should resolve the issue (may take up to 30 minutes to apply).

Last week, I published an article covering the recent Microsoft special event, where they announced Copilot+ PC.

This is a follow-up article, providing an overview of my initial impressions using a Copilot+ PC, specifically the Microsoft Surface Laptop 7.

Although Microsoft is eager to market the Artificial Intelligence (AI) capabilities of a Copilot+ PC, the most interesting thing about this laptop is that it uses the Qualcomm Snapdragon X Elite processor, which is ARM-based (not x86-64).

As highlighted in my previous article, Microsoft has previously attempted to make Windows on ARM viable for consumers. Up to now, every attempt has failed, with poor native ARM performance and inconsistent compatibility with traditional x86-64 software.

With that said, Microsoft are very bullish about the newly announced Qualcomm Snapdragon X Plus and Elite processors, combined with a new x86-64 emulation layer, known as Prism. Therefore, I am excited to see if this laptop can separate itself from past failures and deliver a compelling alternative to the popular ARM-based Apple MacBook Air/Pro series.

The specification of my Surface Laptop 7 can be found below:

Microsoft Surface Laptop 7
Qualcomm Snapdragon X Elite (X1E-80-100 - 12C)
32GB LPDDR5x (8448 MT/s)
1TB M.2 PCI-e 4.0 NVM-e SSD
Wi-Fi 7 and Bluetooth 5.4
15-inch LCD Touchscreen (2496x1664 @ 120Hz)
2x USB-C (USB 4), 1x USB-A (USB 3.1), 1x MicroSDXC, 1x 3.5mm Headphone, 1x Surface Connect

The full specification of the Qualcomm Snapdragon X Elite (X1E-80-100) can be found on the Qualcomm website.

As a reminder, the primary value proposition of an ARM-based laptop is the following.

All day (> 12 hours) battery life for common tasks.
Sustained high performance, even on battery.
Quiet operation, even under load.

This value must be achieved whilst delivering competitive performance across native ARM and traditional x86-64 software.

Hardware

The Surface Laptop 7 is a beautiful laptop, with exceptional build quality. In my opinion, it is the closest alternative to an Apple MacBook, with a minimal design that looks and feels premium.

With this in mind, I will likely use the Apple MacBook as the benchmark throughout this article, as it is clear from Microsoft marketing that they are positioning the Surface Laptop 7 as a direct competitor to Apple’s best-selling laptop (Apple MacBook Air).

The keyboard and trackpad are excellent, delivering a comfortable and familiar experience. The trackpad is not as large as a MacBook, but is haptic, meaning that it can be “clicked” at the top and bottom. It is also multi-touch, but in my experience, it is not as accurate as a MacBook, resulting in occasional unintended swipes, etc. This may be a software limitation, which could improve over time.

The display is also very good, delivering a high resolution (2496x1664) and a 120Hz variable refresh rate, which is an improvement over the MacBook Air (up to 60Hz refresh rate). My only minor complaint is the reflectiveness of the gorilla glass, which can make using the laptop in locations with bright lights a little challenging.

The port selection is also good, with 2x USB-C (USB 4) and 1x USB-A (USB 3.1), alongside MicroSDXC, 3.5mm Headphone and Surface Connect ports. Although the Surface Connect port does offer value (Surface Dock and MagSafe), I would have preferred a HDMI 2.1 port to avoid the need for a dongle when connecting to external displays, etc.

One frustration, the Qualcomm Snapdragon X Plus and Elite processors only support DisplayPort 1.4 (not DisplayPort 2.1). Therefore, the Surface Laptop 7 can not drive my Samsung G95NC Odyssey Neo G9 monitor at its native resolution of 7680x2160. This is a hardware limitation, which likely won’t be improved until the next major revision.

Finally, it is worth mentioning repairability, as although the Surface Laptop 7 is not a Framework Laptop, Microsoft has been taking notes. As highlighted by the iFixit video below, Microsoft has simplified the disassembly process and ensured that all repairable components (e.g., storage, battery) are easily accessible and even labelled with a QR code, providing repair guides, etc.

This has resulted in a very favourable iFixit score of 8/10, highlighting once again that Apple could do better if they prioritised repairability and sustainability.

Emulation

With Windows 11 on ARM 24H2, Microsoft has introduced a new x86-64 emulation engine, known as Prism. Like Rosetta 2 on macOS, the goal is to deliver “acceptable” performance for any non-native (not ARM-based) application.

Unlike Rosetta 2, which is a translation layer, Prism still relies on emulation. Therefore, although Microsoft is touting up to a 3x performance increase, emulation is not as performant as translation.

However, when targetting common workflows (web, productivity, collaboration), the emulation does not need to deliver “native” performance. For example, even a 30% performance reduction could be acceptable, assuming the end-user experience is not negatively impacted.

Rosetta 2 on macOS also takes advantage of the vertical integration of hardware and software that is unique to Apple. For example, the Apple M1/2/3 includes specific hardware features that accelerate x86-64 applications.

Starting with compatibility, I am pleased to report that the majority of my common applications worked as designed (as highlighted in the table below).

The only applications that failed to install were VMware Workstation (not a surprise due to the hardware virtualisation requirements) and the Arc browser, which is expected to have a native ARM version before the end of the year.

Even large, complex, graphics-heavy applications such as Unreal Engine 5 worked and I was able to run the “Third Person Project” at usable frame rates for testing (I am not suggesting the Surface Laptop 7 would be the best laptop for Unreal Engine).

Although I have been impressed with my emulation testing, it is difficult to provide a holistic conclusion, as every user will have a slightly different workflow, requiring a different combination of software. For example, if your workflow requires specialist applications (e.g., graphics, video), I suspect x86-64 will continue to deliver the “best” overall experience (at least for now).

With that said, I was impressed by the number of native ARM applications, which all worked well in my testing, delivering the expected performance based on the hardware specification of the Surface Laptop 7. Specifically, it was great to see common development tools, such as Visual Studio, Visual Studio Code, Docker and Podman all running natively.

Finally, it is worth noting that although many core applications are already ARM native, I noticed that some dependencies (extensions/plugins/agents) used by the core applications are still x86-64. For example, some applications (including Microsoft applications) use a separate agent to manage updates, etc. I do not expect this to cause a major issue, as these dependencies are usually only triggered for specific time-bound tasks and are commonly not performance-limited.

Performance

The table below outlines my high-level performance results for the Surface Laptop 7.

In summary, performance was excellent, often exceeding my expectations, and easily comparable to the x86-64 Intel and AMD alternatives.

Most notably, performance on battery, which is a significant pain point for x86-64 laptops, as they usually require a significant performance drop to achieve acceptable battery life. Thanks to the efficiency of the ARM architecture, the battery performance is consistently excellent and could be the most compelling reason to purchase a Windows on ARM laptop.

Interestingly, I believe this performance could improve over time. For example, I suspect Windows is still (at the core) designed for x86-64 processors that require careful power management to effectively balance performance vs. battery life.

The Surface Laptop 7 includes three “power states”, specifically “Recommended”, “Better Performance” and “Best Performance”. You can see from my testing that when on battery, there is a difference between “Recommended” and “Best Performance”, I assume to further extend the battery life. However, the performance is not consistent in all workloads. Therefore, I suspect Microsoft could continue to “tune” the profiles to further improve the balance. Ideally, these profiles would also automatically trigger based on the workload.

I did not test many games, as the Surface Laptop 7 is not a gaming system. However, I did install Steam and loaded Tomb Raider, which benchmarked well on power and battery, delivering 40+ FPS on “High” settings. Although Tomb Raider is an old (2013) game, the result is still impressive, as Steam and Tomb Raider are not native ARM applications. I did attempt to use the GPD G1 eGPU Docking Station, but unsurprisingly, it was not compatible with Windows on ARM (no native AMD drivers).

Regarding noise, although the Surface Laptop 7 does include active cooling (fans), they are rarely switched on, making the laptop silent for most workloads. This is a major departure from x86-64 laptops (specifically Intel), which frequently sound like a jet engine. With that said, when running stress tests that target multiple parts of the processor (CPU + GPU or CPU + NPU), the fans did engage. At their maximum, the fans do become quite loud (64dB next to the chassis), but this is so rare that I do not consider it an issue.

In summary, the general use of Windows on ARM is very “snappy”, with all applications responding quickly (no obvious stuttering or lag). Although I have no empirical evidence, meaning it could be a case of “the emperor’s new clothes”, Windows on ARM “feels” faster than Windows on x86-64.

Compared to the Apple MacBook Air, my testing highlights that the Apple M3 processor is still the performance and power efficiency king. As a comparison, the Apple MacBook Air M3 achieved Geekbench 6 CPU scores of 3127/12008 and a GPU score of 30541, which is an increase of approximately 10% for the single-core CPU and 40% for the GPU.

This result reinforces the dominance of Apple Silicon, especially acknowledging the MacBook Air with Apple M3 has no active cooling (no fans) and has already been superseded by the Apple M4, which delivers even greater performance.

It should be noted, that the multi-core performance of the Qualcomm Snapdragon X Elite does outperform the Apple MacBook Air with Apple M3 by approximately 19%, thanks to the additional four cores.

With that said, Qualcomm did not need to beat the Apple M3/M4, as this is fundamentally a different platform (Windows vs. macOS). The direct competition for Qualcomm is Intel and AMD, where I am confident they are succeeding. The fact that Qualcomm also delivers a comparable value proposition to the MacBook Air (battery life, performance on battery, silent operation) is a remarkable outcome!

Battery Life

I have not completed any formal battery life testing. However, I have been using the Surface Laptop 7 across a wide range of workloads, switching from power to battery throughout the day.

In my experience, the battery life meets expectations, delivering comfortable “all-day” usage across common productivity and collaboration workloads. The one exception is Microsoft Teams, which can seemingly kill any battery (Windows or macOS) at an alarming rapid pace. Without directly comparing the numbers, the daily usage of the Surface Laptop 7 feels slightly less than what I get with an Apple MacBook, but significantly better than the comparable x86-64 laptops I have tested.

In addition, the standby time of the Surface Laptop 7 is impressive for a Windows-based laptop, dropping by approximately 1% to 3% overnight. In my experience, with x86-64 laptops, this number can be 10% or more.

Enterprise Compatibility

The Surface Laptop 7 was shipped with Windows 11 Home, with no option to pre-install Windows 11 Pro (why, Microsoft). When speaking with Microsoft, they confirmed that the Surface Laptop 7 will be made available to businesses (with Windows 11 Pro) in September.

Thankfully, I was able to upgrade the Surface Laptop 7 using a valid Windows 11 Pro product key. At this point, I could enable business features, as well as onboard the laptop via Microsoft Intune (which automatically transitions the license to Windows 11 Enterprise).

Everything worked as designed, with Microsoft Autopilot, Intune and Defender all running as expected. In addition, I was impressed to see that other enterprise software, such as client VPN, also worked without issue.

The only minor anomaly was a Microsoft Defender compliance issue with Microsoft Intune. However, I suspect this is a short-term problem that will be resolved once the Surface Laptop 7 is officially released for businesses.

Recognising the torrid history of Windows on ARM, it would be bold to transition any business at this time. However, assuming support continues to grow, I could see ARM-based Windows laptops becoming very popular with hybrid workers who spend the majority of their time using web-based and Microsoft 365 (Word, PowerPoint, Excel, Outlook, Teams) applications.

Conclusion

I can confidently state, that the Surface Laptop 7 is the best “all-round” Windows laptop I have ever used, delivering excellent performance and battery life, whilst rivalling Apple when it comes to design and build quality.

The fact that the Surface Laptop 7 outperforms x86-64 counterparts, whilst delivering better battery life and improved thermals is an impressive feat that I believe is a compelling value proposition for anyone considering a Windows laptop.

In addition, it would appear the raw horsepower of the Qualcomm Snapdragon X Elite combined with improved software support (native and emulation) has finally made Windows on ARM not only viable but maybe the preferred platform. It will be very interesting to see how Microsoft positions/manages the x86-64 and ARM versions of Windows moving forward.

With that said, I do not believe Windows on ARM specifically the Qualcomm Snapdragon X Elite is right for everyone (yet). If you are a gamer and/or require specialist applications for your workloads (e.g., graphics, video), I would recommend you stick with x86-64, with a dedicated GPU from NVIDIA/AMD.

Finally, you will note that I did not mention Artificial Intelligence (AI) in this article. At the time of writing, the AI capabilities of Windows and the “benefit” of the Neural Processing Unit (NPU) are essentially non-existent. The flagship feature, Recall, was ironically recalled. The other AI features are not much more than gimmicks that can be easily outperformed by third-party solutions (applications and web services).

In conclusion, I think we have hit a critical milestone for Windows on ARM and ARM-based laptops. Finally, after many years of trying (and failing), Microsoft has delivered on the promise and has a system capable of competing against the Apple MacBook.

Life Sciences AI

Sat, 22 Jun 2024 00:00:00 -0500

I have worked in healthcare and life sciences for the past eighteen years, thirteen years in human health and the past five years in animal health.

Regardless of human or animal, the drug development process is very similar, covering discovery and development, preclinical research, clinical research and approvals. The end-to-end process can take between ten to fifteen years, commonly requiring a significant investment in resources and money.

Therefore, it is fair to state that drug development is considered a very high-risk venture, but if successful, it can deliver world-changing results.

The first phase of this process is arguably the most challenging, where researchers must identify (discover) a specific molecule (e.g., DNA sequence, RNA molecule, protein, metabolite), that plays a role in a disease state and therefore, could be targeted by a drug to produce a therapeutic effect.

Once discovered, researchers search for a compound or compounds that interact with the target molecule and that have the potential to become drug candidates.

A variety of compounds are usually identified, triggering a series of experiments to understand their performance and viability. Typically, this assessment includes factors such as absorption, administration, side effects, and potential interactions. With the completion of these experiments, the most promising compounds are isolated and the preclinical research can begin.

Even from this description, it is clear that the drug discovery and development process is extremely uncertain, and can fail at any point. Therefore, scientists and technologists are continuously looking for new capabilities and/or techniques to improve the chances of success or at minimum, to reduce the time, effort and cost.

Therefore, I thought I would highlight a few companies that are doing interesting things with technology, specifically Artificial Intelligence (AI), to support drug discovery and development.

Insilico Medicine

Insilico Medicine uses AI for target identification, drug discovery, and biomarker development. Their end-to-end AI platform integrates various AI models to accelerate drug discovery and development.

Recursion

Recursion uses AI to analyze cellular imaging data, uncovering new biological insights and drug candidates. Their approach integrates high-throughput experiments with machine learning.

Benevolent

BenevolentAI integrates AI to explore the vast chemical space and identify promising drug candidates. Their platform combines biomedical data with machine learning to discover and develop new treatments.

310 AI

310.ai is building an AI operating system for life sciences, search protein databases, visualise, and run models all in one user-friendly interface.

Exscientia

Exscientia focuses on AI-driven drug discovery and design. Their AI platform accelerates the creation of new drug candidates, optimising properties such as potency and selectivity.

Schrödinger

Schrödinger combines physics-based computational methods with machine learning to design and optimise new drugs. Their platform is used for both small molecule and biologics drug discovery.

Atomwise

Atomwise employs deep learning for structure-based drug design. Their AI technology, AtomNet, is designed to predict the bioactivity of small molecules for drug discovery applications.

Cyclica

Cyclica includes an AI platform, Ligand Express, which predicts polypharmacology and off-target effects of drug candidates, aiding in the design of safer and more effective drugs.

Verge Genomics

Verge Genomics uses AI to map out disease-relevant gene networks and discover drug targets. Their AI platform helps identify novel treatments for neurodegenerative diseases.

Relay Therapeutics (ZebiAI)

Relay Therapeutics uses AI to predict the effects of small molecule drugs on biological targets, helping to identify potential drug candidates faster and more accurately.

Insitro

Insitro applies machine learning to biological data to discover and develop new drugs. Their approach combines data generation at scale with cutting-edge machine learning techniques.

There is no guarantee that these companies will succeed and it is likely many will be acquired as they attempt to sustain their high burn rate through experimentation.

However, I believe it is the convergence of traditional and computational science that will deliver the next wave of innovation within healthcare and life sciences.

Microsoft Copilot+ PC

Fri, 21 Jun 2024 00:00:00 -0500

On the 20th of May, just before their annual Build conference, Microsoft held a special event to formally announce a new generation of Artificial Intelligence (AI) systems, known as Copilot+ PC.

The short promotion video below provides an overview of the announcement.

The special event revealed the following:

New branding and minimum specification for Copilot+ PC.
Windows on ARM with Prism emulation for x86-64 workloads.
Windows Copilot features, specifically Recall and integrations for Paint, Photos, etc.
Qualcomm Snapdragon X Plus and Elite with Copilot+ PC laptops from Microsoft, Dell, Lenovo, Samsung, HP, Asus, and Acer.
New Copilot+ PC options coming from Intel and AMD later this year.

An eligible Copilot+ PC must meet the following minimum specification.

Neural Processing Unit (NPU) with 40 Trillion Operations per Second (TOPS)
8 Logical Processors
16GB Memory
256GB SSD Storage

The most notable requirement is the Neural Processing Unit (NPU), which must be capable of 40 Trillion Operations per Second (TOPS). Currently, only the recently announced Qualcomm Snapdragon X Plus and Elite processors meet this requirement, delivering 45 TOPS. Notably, the Qualcomm Snapdragon X Plus and Elite are ARM-based, not x86-64.

Intel and AMD have announced that they will have Copilot+ PC options later this year, with the release of their next-generation architectures (e.g., Intel Lunar Lake and AMD Zen 5).

The requirement for a dedicated NPU is interesting, as it excludes other capable hardware, such as a Graphics Processing Unit (GPU). For example, a mid-range NVIDIA GeForce 4050 Laptop GPU can deliver up to 194 TOPS, which is significantly above the 40 TOPS requirement. However, as it is not a dedicated NPU, it does not meet the requirement, meaning it will not gain access to Copilot+ PC features.

It is fair to state, that a dedicated GPU would not be as power-efficient as an NPU, but power efficiency is not always a requirement, specifically for workstation-class laptops or desktops, which are predominately mains-powered.

Therefore, the dedicated NPU requirement does feel like a barrier positioned by Microsoft marketing to clearly differentiate and force new hardware purchases and/or upgrades (which is a bit frustrating). Microsoft leveraged a similar tactic as part of the Windows 11 release, requiring specific processors and TPM 2.0.

The memory and storage requirements make more sense, as the use of local AI capabilities, such as large language models (LLMs), can be resource-heavy. For example, Microsoft has stated they plan to ship 40+ large language models with Windows (stored and running locally) and the newly announced Recall feature requires a minimum of 50GB storage.

Windows AI

At the event, a lot of emphasis was placed on the Windows AI features included with Copilot+ PC, targeting Windows 11 24H2 (previously rumoured to be Windows 12). Unfortunately, since the announcement, there has been a strong consumer response to the headline feature, known as Recall, highlighting privacy and security concerns.

Recall is an explorable timeline, which works by capturing a screenshot automatically every five seconds. These screenshots are then made available to a local large language model, which can be used to query (recall) anything the user previously worked on via natural language. For example, “Show me the slide I was working on which included a picture of a red dragon”.

The feature itself is quite interesting, as the human mind commonly places greater emphasis on visual cues vs. file systems when attempting to retrace memories.

Unfortunately, and you would assume, fairly obviously, that any feature that takes a screenshot every five seconds would pose a legitimate privacy and security concern. For example, post-announcement, the cybersecurity community demonstrated examples of malware that could target and copy Recall data, making it a convenient target for cybercriminals.

As a result, after several rounds of deliberation, Microsoft has formally delayed Recall. Therefore, it will not be available at launch with the first round of Copilot+ PC laptops, providing time for them to verify and improve the privacy/security controls.

Windows on ARM

Although the Windows AI features took centre stage, I was personally more interested in the switch from x86-64 to ARM, with the new Qualcomm Snapdragon X Plus and Elite processors.

Microsoft has attempted (many times) to make Windows on ARM viable for consumers. Up to now, every attempt has resulted in failure, with poor native ARM performance and inconsistent compatibility with traditional x86-64 software. Therefore, unsurprisingly, PC manufacturers and the developer community have largely ignored Windows on ARM.

This has been a point of frustration for Windows users, as the x86-64 market with Intel and AMD has been stagnant for many years, with uninspired incremental hardware revisions. It also placed the PC market at a disadvantage against Apple and their Apple Silicon laptops/desktops, which successfully switched to ARM in 2020.

To be clear, there is nothing inherently wrong with the x86-64 architecture. However, the ARM architecture has proven very capable in three key areas.

All day (> 20 hours) battery life for common tasks.
Sustained high performance, even on battery.
Quiet operation, even under load.

These three areas are well suited to the ARM architecture, which is why it dominates the smartphone and tablet market. The Apple MacBook Air/Pro line of laptops has also proven the viability for larger computing devices, assuming the performance and software compatibility exists.

In short, the PC market has failed to deliver a compelling Apple MacBook Air competitor with x86-64. Even when they have come close from a form-factor perspective, the performance on the battery (when not powered) is dramatically reduced, as x86-64 processors must throttle aggressively to maintain reasonable battery life.

Thankfully, there is growing optimism that the newly announced Qualcomm Snapdragon X Plus and Elite processors, combined with the new Prism emulation layer will change this story, delivering the required combination of performance and compatibility to make Windows on ARM viable at scale.

Initial signs are positive, with encouraging commitment from PC manufacturers (Dell, Lenovo, Samsung, HP, Asus, Acer) and the developer community (Google, Adobe, etc.)

Microsoft Surface Laptop 7

Earlier today, I received a Microsoft Surface Laptop 7 for testing, which is a Copilot+ PC laptop, running Windows on ARM via a Qualcomm Snapdragon X Elite processor.

A summary of specification can be found below:

Microsoft Surface Laptop 7
Qualcomm Snapdragon X Elite (X1E-80-100 - 12C)
32GB LPDDR5x (8448 MT/s)
1TB M.2 PCI-e 4.0 NVM-e SSD
Wi-Fi 7 and Bluetooth 5.4
15-inch LCD Touchscreen (2496x1664 @ 120Hz)
2x USB-C (USB 4), 1x USB-A (USB 3.1), 1x MicroSDXC, 1x 3.5mm Headphone, 1x Surface Connect

The full specification of the Qualcomm Snapdragon X Elite (X1E-80-100) can be found on the Qualcomm website.

Over the next week, I will be testing the Microsoft Surface Laptop 7. Therefore, stay tuned for a follow-up article with the results. I am excited to see if ARM is the future of Windows!

Threat Intelligence Podcast

Thu, 20 Jun 2024 00:00:00 -0500

In March, I was a guest on the Future of Threat Intelligence podcast, hosted by David Monnier (CIO - Team Cymru).

The episode has now been published, links below:

The Future of Threat Intelligence podcast is dedicated to helping security experts and leading practitioners succeed in cybersecurity and risk management.

In the episode, we discuss the unique challenges and opportunities of the life sciences industry and experience from our recent corporate separation. We emphasise the importance of practical application and hands-on experience in cybersecurity.

We also talk about the necessity for proactive measures to tackle evolving threats. This includes the emergence of passwordless (FIDO2), Generative AI and modern automation techniques.

It was a great discussion, thanks Team Cymru!

Generative AI Value

Sat, 15 Jun 2024 00:00:00 -0500

I have documented this journey via a series of articles (links below), including our recent external recognition at the BTOES Awards.

Since my last update in December, we have continued to see viral growth and sustained adoption. The screenshot below highlights our real-time dashboard.

Since launch, 6,144 unique users have accessed the service, which is 88% of the target audience. We see approximately 1,600 interactions per day, which is an approximate increase of 60% since December.

Our custom persona capability, which I outlined in my last article has been used to create 683 business-specific Generative AI chatbots, which can be shared via our internal marketplace.

As a result, we predict 27,317 minutes have been saved, resulting in an estimated return on investment of $1,308,616, driving through productivity acceleration and cost avoidance.

In addition, the Generative AI framework has been used to streamline and/or evolve specific business processes. Outlined below are six examples of business opportunities and their outcomes. All examples deliver a unique return on investment, above and beyond the value associated with productivity acceleration and cost avoidance.

AskSAP - Customer Orders

Business Opportunity: Replacing a time-consuming manual process regarding customer order queries, which was triggered by sales representatives ~17,000 in 2023, costing an estimated $250,000.
Outcome: Real-time access to customer order information from SAP, including advanced queries to unlock customer insights.

AskSAP - Purchase Orders

Business Opportunity: Dramatic reduction in user effort and complexity accessing Purchase Order and Invoice information, increasing efficiency and improving compliance.
Outcome: Real-time access to Purchase Orders, Invoices and by extension, vendor expectations.

Warehouse Management

Business Opportunity: Warehouse engagement via email commonly lacks the required context to map specific requests to SKUs, resulting in order delays and/or errors, impacting customer satisfaction and inventory management.
Outcome: Corporate B2B customer ordering via a controlled chatbot, replacing the legacy email process.

Manufacturing Quality Advisor

Business Opportunity: Manufacturing sites must continuously maintain compliance, with any issues, potentially resulting in production stoppages, infraction letters, etc. The complexity of the site operations, documented across many policies and processes can contribute to errors, impacting both productivity and profitability. An estimated $1,300,000 is associated directly with execution errors.
Outcome: Leverage language capabilities to automatically sort, summarise, compare and re-structure 2800 documents.

Pharmacovigilance

Business Opportunity: A contact centre agent must accurately gather all necessary information, extracting key information to ensure the appropriate and timely next steps. Approximately 9000 cases a month (82% new).
Outcome: Automated extraction and summarisation of key information, structured consistently to ensure accuracy, whilst reducing the burden on the agent by simplifying the input process and case management.

Document Translations

Business Opportunity: Over $500,000 is spent on language translations per year, leveraging external agencies. In many cases, content is not translated due to time/budget constraints, which impacts adoption.
Outcome: Automated document translation with domain-specific context and expertise, covering 135 languages, translating up to 40 languages simultaneously.

These examples are just a small sample, with new opportunities being identified and implemented every week!

Project Starline

Tue, 11 Jun 2024 00:00:00 -0500

Earlier today, I tested Google Project Starline, which aims to deliver an immersive telepresence experience, where the participants are life-sized and displayed in three dimensions, without the need to wear glasses or a headset.

As described by Google, Project Starline is like looking through a magic window, where you can see another person, with volume and depth. You can talk naturally, gesture and make eye contact. The effect is the feeling of a person sitting directly in front of you.

Project Starline combines computer vision, machine learning, spatial audio, a light field display and real-time data compression.

The latest version of Project Starline (which has been through multiple iterations) is delivered as a compact series of cameras, attached to what appears to be a regular display.

In use, the combination of spatial audio and the light field display successfully tricks the mind into believing the individual is present within the room, delivering an impressively realistic effect.

Unfortunately, I was not able to record my test session, but even if I had, the “magic” would not translate to video. This is due to the use of eye-tracking technology, which is what maintains the visual illusion. Therefore, Project Skyline can only be experienced in person.

The best overview/description I have seen is the Marques Brownlee (MKBHD) video from last year.

I completed a similar series of tests as Marques (e.g., passing the apple). My reaction was similar to his, initially, a feeling of disbelief, followed by a desire to see how far the technology could be pushed before the illusion was broken. In the end, I settled into a natural conversation with the other person, which was the most realistic I have ever experienced via virtual conferencing.

The specifics of the technology were left a little vague. For example, I would love to know exactly how the light field display works. Without any insider knowledge, Looking Glass appears to be the closest comparison.

Information gleaned from other interviews (e.g., Wired) states that the depth sensors are capable of capturing approximately 180 degrees. Therefore, if you move outside of this window, the volume and depth effects are lost.

The data compression technology leverages a modified version of WebRTC, which is rumoured to reduce the size of the raw data by a factor of 100. This is what enables Project Starline to operate over a traditional network (although it would appear all the test sessions are connected within the same building).

At this time, Project Starline remains a “Technology Project” and has specific restrictions. For example, it only supports one-to-one discussions. However, Google shared that they plan to commercialise the product in 2025, working with partners such as HP.

The real-world applicability of Project Starline is yet to be proven. However, I can certainly see it being popular with business executives.

Looking beyond 2025, I am interested in understanding how quickly this technology could be delivered to a wider audience, as it certainly feels like a positive step forward from traditional virtual conferencing technologies.

Framework Paused

Tue, 04 Jun 2024 00:00:00 -0500

As highlighted in my last article, I have been partnering with Framework to bring sustainable computing to enterprise businesses, starting with laptops.

I have documented this journey, including my own experience working with Framework, across a series of articles.

With over fifteen years of experience working with traditional laptop manufacturers (e.g., Dell, Lenovo, HP), I have long felt the life cycle management and servicing business models were fundamentally flawed, with carefully engineered processes that favour manufacturers, requiring consumers and companies to buy new devices on a regular cadence (commonly three to five years).

Therefore, my team and I set ourselves an ambitious goal to redefine these legacy business models, delivering an alternative approach that prioritises environmental sustainability and the right to repair. In our case, we predict the potential for an estimated 33% reduction in e-waste, alongside a positive return on investment for the business, saving an estimated millions over a ten-year lifecycle.

In the article “Framework Business Process”, I outlined our approach, including principles, processes and technical details.

Earlier this year, we launched a business preview, deploying approximately 100 laptops, across the UK, US and Germany. The majority are Framework Laptop 13 (a mix of Intel and AMD), alongside a few Framework Laptop 16.

The business preview itself was a success, verifying the viability of the concept. However, due to specific reliability and support findings, the value was not achieved, resulting in the decision to temporarily pause any additional scale.

In the spirit of transparency, outlined below are the top findings identified as part of the business preview.

Reliability

Boot Device Not Found: ~75% failure rate with the Framework Laptop 13 AMD Ryzen 7040 Series, where the boot device becomes unreadable, requiring a full reinstallation of Windows 11.
Trackpad: ~10% failure rate with the Framework Laptop 13 (AMD and Intel), where the trackpad remains functional but no longer “clicks” as designed.
Device Management Compliance: ~100% failure rate with the Framework Laptop 13 AMD Ryzen 7040 Series, where the Pluton TPM fails attestation within Microsoft Intune.
Dead on Arrival: 5% failure rate with the Framework Laptop 13 Intel Series, where the displays arrived (out of the box) cracked.

The failure rates were certainly higher than anticipated and above what we have seen with equivalent Dell/Lenovo laptops.

We also noticed the impact of “wear and tear” was high, resulting in an increased failure rate for mechanical parts (e.g., trackpad).

In all cases, the laptop could be recovered (software reset or hardware replacement). However, these activities take time and effort, impacting the return on investment.

At this stage, it is unclear if these finders are isolated to our business preview or evidence that business users are “harder” on their devices, increasing the rate of failure with the modular nature of a Framework laptop.

Support

Technical Support: Support tickets directed by our business account manager at Framework took many weeks/months to resolve. This included the coordination of replacement parts and the bugs identified in the software stack (specifically the AMD firmware).
Shipping Quality: The delivery of replacement parts was frequently delayed or damaged in transit. Specifically, multiple failed attempts to deliver to our offices in Germany due to failed hand-offs between carriers and import restrictions.
Invoicing: At the time of writing, Framework does not have established relationships with Value-Added Resellers (VARs), which are commonly used by enterprise businesses to enable global scale. As a result, direct purchasing was required, with issues associated with the timeliness of invoices, impacting internal financial processes.

The highlighted support findings were not unexpected based on the relative immaturity of the company and business processes. It should also be noted, that the Framework team were actively engaged and demonstrated a genuine desire to support and improve. This included regular meetings with Nirav (Founder and CEO of Framework). Therefore, I expect some of these issues to be resolved organically over time, as the company matures.

Finally, in addition to the reliability and support findings, we have also been monitoring the market closely, specifically, the turbulence being generated by Microsoft regarding “AI PC” capabilities and the anticipated launch of ARM-based processors from Qualcomm. I believe these market dynamics could trigger a wave of innovation from Intel/AMD, making any buying decision in 2024 difficult.

These external factors, contributed to our decision to pause, providing time to reflect, learn and improve with Framework, whilst also allowing the market to stabilise.

Framework GDP G1

Tue, 21 May 2024 00:00:00 -0500

In late 2022, I initiated a partnership with Framework to bring sustainable computing to enterprise businesses, starting with laptops.

I have documented this journey, including my own experience working with Framework, across the following articles.

In the article “Framework Business Process”, I outlined our approach, including principles, processes, technical details, etc.

The majority of our testing was with the Framework Laptop 13, which covers 100+ users in production across the UK, Germany, and the US.

Although the Framework Laptop 13 is a great laptop that delivers a good balance of portability and performance, it does not include an option for discrete graphics (e.g., NVIDIA or AMD). This is viable on the Framework Laptop 16. However, as per my article “Framework Decision”, the compromise regarding size, weight and build quality is significant.

Therefore, we have positioned the use of an eGPU for the limited number of users that require additional graphics horsepower, either to run specialist applications and/or to grant access to computing platforms such as NVIDIA CUDA.

This approach does limit the overall performance of the discrete graphics card, commonly achieving between 70% to 80% of the total performance. However, it is highly versatile, as the eGPU can be easily shared across teams and customisable/upgradable without restriction.

As part of our testing, we purchased the GDP G1 eGPU Docking Station, which is not the cheapest (£686.95) or most powerful eGPU on the market, but is compact and includes additional ports.

The full specification can be found below:

AMD Radeon RX 7600M XT (8GB GDDR6)
1x Oculink（SFF-8612), 1x USB 4
3x USB 3.2, 1x HDMI 2.1, 2x DP 1.4a, 1x SD Reader 4.0

The AMD Radeon RX 7600M XT is built on the 6nm RDNA3 architecture, with 4096 stream processors, 32MB infinity cache and 13.3B transistors.

When connected via USB 4, it delivers 63Gb/s, alongside 65W power to the laptop, including fast charging via PD protocol.

Recognising the target audience, the testing was completed using Linux, specifically Ubuntu, which is our corporate standard.

The first screenshot below highlights the default Framework Laptop 13 Geekbench 6 score, using the Vulkan API. This includes the AMD Radeon 780M, with 12 graphics cores running at 2.7GHz.

When connected via USB 4, the GDP G1 eGPU Docking Station is made immediately available for use (screenshot below). Thanks to the excellent support for AMD drivers within Linux, no additional software or configuration is required (plug and play).

The GDP G1 eGPU Docking Station includes two Total Graphics Power (TGP) modes, toggled via a button on the front of the unit. The first is the quiet mode, delivering 60W TGP, and the second is the normal mode, delivering 100W TGP. The screenshot below highlights the Geekbench 6 score, using quiet mode (60W TGP).

The screenshot below is the GDP G1 eGPU Docking Station running in normal mode (100W), highlighting the maximum obtainable performance.

The performance improvement from the integrated AMD Radeon 780M to the discrete AMD Radeon RX 7600M XT is approximately 66%, when running in normal mode (100W). Although this mode is in theory louder due to the higher TGP, the actual fan noise is perfectly acceptable (no louder than most gaming laptops).

Overall, the eGPU option worked very well, delivering a highly versatile method of increasing graphics performance, whilst maintaining the modular/sustainable philosophy promoted by Framework.

The only minor issue identified during the testing was a bug during boot when full disk encryption was enabled for Ubuntu. The system would simply hang until the eGPU was switched off or physically disconnected. At this point, the system booted as expected.

Next CIO Awards UK

Thu, 16 May 2024 00:00:00 -0500

Thanks to CIO.com for the recognition, being named a Next CIO Awards UK 2024 winner, which aims to “recognising the rising stars of the UK IT sector”.

CIO.com is a well-established publication, that attracts the highest concentration of enterprise business technology executives, offering unparalleled peer insights and expertise on business strategy, innovation, and leadership.

Next CIO highlights digital, data and technology leaders, joining an exclusive community that supports inclusion, career development and progress. It is also an opportunity for current/existing CIOs and business leaders to reward their colleagues and peers for what they have achieved.

The 2024 judging panel included representation from: Microsoft, University of London, Stantec, United Living, Datapharm, etc.

I am proud to represent my amazing team, as we partner to provide those who raise and care for animals with products/services that empower them to solve some of the world’s most pressing issues. Congratulations to all the winners!

Common PIN Codes

Wed, 15 May 2024 00:00:00 -0500

Earlier this week, the following chart was published/shared via Reddit, highlighting the most common four-digit PIN codes. The source of the analysis was a blog post from DataGenetics.

The analysis leverages 3.4 million PIN codes, collated from released/exposed/discovered credential tables and public security breaches.

The data highlights that the top twenty PIN codes constitute 27% of all PIN codes, with common themes being two pairs and birth years.

The reason this information is valuable is that PIN codes are becoming more relevant as consumers and businesses look to embrace Windows Hello for passwordless authentication.

Windows Hello is a significant step forward for authentication, as it is FIDO2 compliant, which is phishing-resistant. By default, Windows Hello leverages biometrics (facial recognition or fingerprint), but is also backed by a PIN code (minimum four digits).

On the surface, the use of a PIN code feels less secure than a complex password. However, the key difference is the FIDO2 compliance and the fact the PIN code is only associated with one device (desktop, laptop, smartphone, etc.) Therefore, to use the PIN code, you must have physical access to the device itself (in theory, it cannot be compromised remotely).

In addition, when using Windows Hello, the PIN code is stored in the Trusted Platform Module (TPM), which includes anti-hammering protection, which aims to prevent the PIN code from being brute forced.

By default, the TPM (v2.0) is configured by Windows to lock after 32 authorisation failures and to forget one authorisation failure every 10 minutes. This configuration can be customised, which should be considered based on the risk profile of the business/user/data. Additional information can be found in the TPM Fundamentals article published by Microsoft.

Assuming a four-digit PIN code is used, with the default TPM anti-hammering protection, attempting all 9999 possible PIN combinations would take approximately two years.

With that said, I would still recommend leveraging the insights from this data analysis to select a “strong” PIN code (minimum six digits), targeting a unique number sequence.

BTOES Awards

Fri, 10 May 2024 00:00:00 -0500

This week, my team attended The Business Transformation & Operational Excellence Summit (BTOES) industry awards in Orlando, where we were selected as a finalist in the category of “Best achievement in deployment of AI in the enterprise”.

The team were recognised for their work on ElancoGPT, a multi-model, multi-modal Generative AI framework, designed to accelerate productivity by streamlining common business tasks.

ElancoGPT is private to Elanco, with specific business features and controls to ensure security, privacy, legal, and compliance. It leverages Retrieval-Augmented Generation (RAG) to ingest Elanco data, enhancing the accuracy and reliability of output.

I have documented the design and development of ElancoGPT via a series of blog posts.

Since launch, ElancoGPT has seen dramatic organic adoption, covering ~63% of the FTE workforce, generating an estimated ROI of $1,063,344 through productivity acceleration.

A summary of the 2024 finalists and winners can be found on the BTOS website

Early next month, I plan to provide an update on our progress, highlighting our latest features, roadmap and learnings.

3DFX Voodoo Banshee

Sat, 27 Apr 2024 00:00:00 -0500

My passion for computing formally started in 1992 (I was 8), when my parents purchased our first PC and my dad and I successfully upgraded the processor from an Intel i486SX to an Intel i486 OverDrive.

This moment began a lifelong love of computer hardware and ultimately, all things technology (e.g., hardware, software, data, networking, gadgets, etc.)

However, it was in 1998 (I was 14), when my parents bought me my first 3D accelerated graphics card, that things really changed.

Therefore, to this day, 25 years later, I still consider the 3DFX Voodoo Banshee PCI 16MB the most important and influential technology purchase of my life.

This upgrade opened my eyes to an entirely new world of 3D accelerated games, including Quake (1996), specifically GLQuake (1997), Half Life (1998), Unreal (1998), etc.

Up until this point, I had relied upon software rendering within games, which inherently restricted the resolution, frame rate and graphics techniques, such as transparency, shadows, etc.

I vividly remember loading Half Life and watching the opening sequence in full 3D accelerated glory and being blown away. I spent the next decade building, modifying, and tweaking computers, looking for any opportunity to push the hardware and software to its limits.

The 3DFX Voodoo Banshee was an interesting graphics card, as it was a single-card, single-chip solution, combining 2D and 3D acceleration. This was the first time 3DFX had released a single-chip solution, following the commercial failure of the 3DFX Voodoo Rush, which was a single-card solution but accelerated 2D and 3D graphics via different chips.

By comparison, the original 3DFX Voodoo was a dedicated 3D accelerator card, which required an additional video controller for 2D acceleration. The implementation required a pass-through VGA cable daisy-chained, which increased the complexity and cost for the consumer.

Therefore, combining 2D and 3D acceleration onto a single card and a single chip was the future.

Interestingly, you would assume, based on the history of the 3DFX, that the 3D acceleration of the 3DFX Voodoo Banshee would be the highlight. However, it was the 2D acceleration that made the product special. For example, all Windows GDI functions were implemented in hardware, with a 128Bit 2D core matching VESA 3.0 specifications, making the 3DFX Voodoo Banshee one of the fastest 2D accelerators on the market and an excellent choice for DOS gaming.

Ironically, the 3D acceleration was a little more controversial, as although it supported Microsoft DirectX 6.0 and OpenGL 1.1, the 3DFX Voodoo Banshee was equipped with a single texture unit, unlike the 3DFX Voodoo 2, which had two texture units. As a result, the 3DFX Voodoo Banshee was slower when processing multi-textured environments. Therefore, anyone looking for the “best” 3D acceleration from a 3DFX graphics card would be better served by the 3DFX Voodoo 2.

From memory, my dad and I selected the 3DFX Voodoo Banshee as it was a cost-effective, single-card 2D/3D accelerator, which was compatible with PCI (vs. AGP).

In summary, even with industry defining products such as the Apple iPhone, I will always look back with memories of admiration and excitement regarding the 3DFX Voodoo Banshee.

NOTE: At the time, I did not recognise the importance of this card to me personally. Therefore, as is often the case, it was lost to time. However, I recently purchased one on eBay and framed it in my office.

It is common for people who start their own business to frame their first note earned. This is my version of that tradition, the graphics card that started my LifeinTECH.

Ubuntu Cheat Sheet

Thu, 25 Apr 2024 00:00:00 -0500

Earlier today, Canonical released Ubuntu Desktop 24.04 LTS Noble Numbat. I recommend reviewing their recent blog post for a summary of the new features, etc.

NixOS and Fedora are my preferred Linux distributions. However, at work, I still use Ubuntu Desktop thanks to the stronger support for Microsoft Endpoint Management (Intune) and Protection (Defender).

Therefore, I thought I would share my initial setup steps (which can be automated) for Ubuntu Desktop.

Ubuntu Desktop Configuration

Ubuntu Desktop 24.04 LTS Noble Numbat can be downloaded from the Canonical website.

I create a bootable USB using balenaEtcher, which supports Windows, macOS and Linux.

As part of the initial Ubuntu Desktop installation, I configure the following:

Secure Configuration:

Full Disk Encryption: LVM and Encryption
Encryption Passphrase: Minimum 12 Characters
Login Password: Minimum 12 Characters

I also install the following corporate endpoint management and protection software.

Security Software:

Ubuntu Pro Configuration

The following commands enable Ubuntu Pro, which is a low-cost (free for personal use) subscription service that aims to provide stability, security, and compliance for Ubuntu installations.

sudo pro attach ...  
pro --version  
pro security-status  
pro security-status --esm-apps

NOTE: The attach command requires your unique token, which can be found on the Ubuntu Pro Dashboard.

Ubuntu Pro includes the following useful features.

Expanded Security Maintenance (ESM): ESM provides 10 years of vulnerability management for critical, high and selected medium CVEs for all software packages shipped with Ubuntu.
Live Kernel Updates: Minimise downtime and unplanned reboots with patches for critical and high-severity kernel vulnerabilities.
Compliance and Hardening: Certified automation tools for hardening and compliance profiles. Specifically, I target “cis_level1_workstation”.

Ubuntu Package Manager

The following commands are useful when interacting with the Ubuntu Advanced Packaging Tool (APT).

sudo apt update  
sudo apt upgrade  
sudo apt install ...  
sudo apt remove ...  
sudo apt purge ...  
sudo apt download ...  
sudo add-apt-repository -L  
sudo add-apt-repository ...

I would always recommend running ‘sudo apt update’ and ‘sudo apt upgrade’ following a new installation to ensure the operating system and software packages are up to date.

Recommended Snaps (Applications)

Although I prefer Flatpak, Canonical continues to favour snaps. Therefore, outlined below are the applications I install via snap, all of which are verified.

sudo snap install libreoffice  
sudo snap install slack  
sudo snap install code --classic  
sudo snap install firefox  
sudo snap install standard-notes  
sudo snap install bitwarden  
sudo snap install gimp  
sudo snap install krita  
sudo snap install audacity  
sudo snap install vlc  
sudo snap install natron  
sudo snap install obs-studio  
sudo snap install gnome-boxes  
sudo snap install postman  
sudo snap install godot-4  
sudo snap install steam

I use Obsidian for note-taking, which can also be installed via a snap. However, it must be downloaded manually from the Obsidian website and installed using the following command.

snap install obsidian_1.5.12_amd64.snap --dangerous --classic

The “dangerous” flag sounds scary, but is OK when the snap is downloaded from a trusted source. In short, it acknowledges that Canonical has not reviewed this snap.

Finally, the following commands are useful when interacting with snaps.

snap refresh --list
snap info ...
sudo snap refresh
sudo snap refresh ...

The ‘sudo snap refresh’ command will update all installed snaps.

Recommended APT Packages (Applications)

The following applications are not available, viable and/or verified as a snap. As a result, I install them using the Ubuntu Advanced Packaging Tool (APT).

The following command installs the common Microsoft fonts.

sudo apt install ttf-mscorefonts-installer

The following command installs and starts Docker and Docker Compose.

sudo apt-get install ca-certificates curl  
sudo install -m 0755 -d /etc/apt/keyrings  
sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc  
sudo chmod a+r /etc/apt/keyrings/docker.asc  
echo \  
	  "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu \  
	  $(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \  
	  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null  
sudo apt-get update  
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin  
sudo service docker start

At work, we use Microsoft 365 for productivity, collaboration and security. As Microsoft do not have native applications for Linux, I would recommend installing Microsoft Edge for improved compatibility with their web services. To install, you must first download Microsoft Edge from the Microsoft website.

It can then be installed using the following command.

sudo apt install ./microsoft-edge-stable_124.0.2478.51-1_amd64.deb

Openshot, which is a popular open-source video editing application, can be installed via the following commands.

sudo add-apt-repository ppa:openshot.developers/ppa  
sudo apt update  
sudo apt install openshot-qt python3-openshot

Quickemu, is a great virtualisation application that automates the download, installation and configuration of virtual machines (Windows, macOS, Linux, etc.) The following commands install Quickemu.

sudo apt-add-repository ppa:flexiondotorg/quickemu
sudo apt update
sudo apt install quickemu

Finally, Ollama is an application for running local Large Language Models (LLMs), including popular open-source options such as Meta Llama 3, etc. The following command installs Ollama.

curl -fsSL https://ollama.com/install.sh | sh

That’s it! Ubuntu Desktop is now ready for work. There are other applications that I install, but I consider these personal/optional, which I tend to install as needed and not as part of my initial setup.

TechPoint MIRA Awards 2024

Wed, 24 Apr 2024 00:00:00 -0500

Update - Unfortunately, the team did not win the main prize. However, it was great to engage the community, celebrating the thought leadership and innovation. Congratulations to all the winners.

For the third consecutive year, my team has been nominated and selected as a winner for a TechPoint MIRA award.

In 2024, the team were highlighted in the category of “Digital Transformation”, which recognises teams from any industry that have demonstrated outstanding vision and innovation for driving digital adoption and leveraging technology to revolutionise and elevate their operations and/or overall industry standards.

Our submission outlined our new omnichannel digital engagement strategy, which successfully increased customer touch points by 30%, reducing the average time to first purchase by 28%. In a pilot with 10,000 customers, we also observed a 4% increase in sales.

The short video below highlights the TechPoint MIRA Awards 2023 gala, which is the annual event where the final winners are announced.

The winners of 2024 TechPoint MIRA Awards will be announced later this week on Friday, 26 April 2024. Wish us luck!

SXSW 2024

Fri, 19 Apr 2024 00:00:00 -0500

As highlighted in previous articles, at my company, we prioritise our “Healthy Purpose”, which is the platform through which we engage with our employees, communities and industry leaders across more than 90 countries, to provide global, sustainable solutions that make an impact.

As CTO/CISO, I look to ensure IT is acting as a thought leader, exploring new ways to minimise our environmental footprint, and prioritising partnerships that unlock innovation to advance our sustainability efforts. Our most recent initiative involves a unique praetorship with the company Framework to bring sustainable computing to enterprise businesses, starting with laptops.

In parallel my CEO and mentor, Jeff Simmons, recently presented alongside a panel of experts (Frank Mitloehner - CLEAR Center and Kendra Tolley - Athian) at South by Southwest (SXSW) 2024, focused on sustainable farming.

Dairy and livestock production have long been named “climate culprits.” However, we believe that meat and dairy need to be part of the climate solution. The SXSW presentation provides insights into how reducing methane emissions in livestock can lead to environmental sustainability improvements while feeding a growing world population high-quality protein.

Jeff and the panel unpack how technologies and innovations are revolutionising agriculture putting climate-neutral cattle farming within reach this decade.

The session is over one hour, but well worth the time. I believe the panel did a nice job of presenting the data, providing the required context to better understand this complex topic, whilst outlining a science-based, pragmatic path forward.

Nirav Patel

Tue, 16 Apr 2024 00:00:00 -0500

As highlighted in my previous articles, my team has been partnering with Framework to bring sustainable computing to enterprise businesses, starting with laptops.

Earlier this month, following an insightful pilot, we launched our business preview in Europe. Initial feedback has been very positive and I plan to share more details soon.

Complimenting this initiative, Nirav Patel (Founder and CEO of Framework) recently completed an interview with Jason Carman, who has an excellent YouTube series called S3 focused on startups.

Therefore, if you watch one video this week, I would recommend the interview with Nirav. He speaks to his approach to innovation, following his experience building FaceTime at Apple and becoming the first employee at Oculus, before the Meta (Facebook) acquisition.

A few snippets (not quotes) that resonate strongly with me.

“Follow your passion and stay curious, seek to understand the “why” by getting hands-on”.

“Never try to replicate another company or product. Chart your own course, start lean, iterate quickly and continuously learn”.

“Use your own products early. Even when they are flaky”.

“Target markets where your ideas force competitors to challenge their traditional business models”.

“Establish partnerships based on an aligned sense of purpose and values”.

The interview is full of fascinating insights, many of which are behaviours/approaches that I personally foster and openly promote.

Over the past year, my team has built a great relationship with Nirav and I can confirm he is one of the hardest-working individuals I have ever met. Therefore, I have a lot of respect for his insights.

Over the coming months, I intend to share more details regarding our partnership and scale (we expect to deploy ~1600 Framework laptops in 2024), alongside some official collaborations with the Framework team. Exciting times ahead!

Future of Threat Intelligence

Thu, 28 Mar 2024 00:00:00 -0500

Update - The episode is now live.

Earlier today, I joined David Monnier (CIO - Team Cymru) as a guest on the Future of Threat Intelligence podcast, also available on Apple Podcasts.

The Future of Threat Intelligence podcast is dedicated to helping security experts and leading practitioners succeed in cybersecurity and risk management. Each episode highlights actionable takeaways to help the audience get ahead of the curve and prepare for the trends and technologies shaping the future.

It was a great discussion, covering a wide range of topics, specifically my background/career, threat intelligence, cyber risk strategy, emerging technologies, talent development, and recommendations.

The podcast should be published in the next couple of months.

Framework Business Launch

Mon, 25 Mar 2024 00:00:00 -0500

In late 2022, I initiated a pioneering partnership with Framework, to bring sustainable computing to enterprise businesses, starting with laptops.

With over fifteen years of experience working with traditional laptop manufacturers, I have long felt the business models supporting life cycle management and servicing were fundamentally broken, favouring the manufacturer that forces consumers and companies to buy new devices regularly.

In 2024, as Chief Technology Officer (CTO) at my company, we are taking a stand against this practice, which we believe will reduce our e-waste by ~33%, whilst also delivering a positive return on investment for the business, saving an estimated $18M over a ten-year lifecycle.

I have documented this journey, covering my personal and business experience, over a series of articles. Specifically, the articles “Framework for Business Pilot” and “Framework Business Process”.

Last week, Framework formally acknowledged our progress with the official launch of “Framework for Business”, which now has a prominent position on their home page.

Although still very early, the Framework for Business webpage provides useful context, alongside key information for businesses looking to explore sustainable computing. I am also excited to see our endorsement and support clearly positioned.

Over the coming months, I intend to share more details regarding our partnership and scale (we expect to deploy ~1600 laptops in 2024), alongside some official collaborations with the Framework team. Exciting times ahead!

Framework Decision

Sun, 10 Mar 2024 00:00:00 -0600

Since January, I have been testing the Framework Laptop 16.

As highlighted in my initial thoughts article, my overall impression has been positive and I continue to commend the Framework team on their design/manufacturing expertise and dedication to sustainability.

With that said, the Framework Laptop 16 is not perfect, with a few areas that I believe need improvement. Thankfully, the serviceable/upgradable nature of Framework products allows for improvements post-release, something that the Framework team demonstrated very effectively with the Framework Laptop 13.

For example, I would like to see an option for a full-width keyboard and trackpad input cover, alongside a higher-quality webcam, microphone and speakers.

Therefore, I fully expect the Framework Laptop 16 to be a “better” product one year from now, which is rare with modern electronics, as manufacturers frequently fail to offer comprehensive post-release support.

Unfortunately, even with this future potential, I have decided to switch back to the Framework Laptop 13.

The primary reason is the size and weight of the laptop, which is especially noticeable when the Graphics Module is installed. As a comparison, the photo below highlights the Framework Laptop 16 under an Apple MacBook Pro 16-inch.

Without the Graphics Module, the laptop becomes more manageable. However, you lose the benefits of a dedicated GPU, which in my testing is the primary performance differentiator between the Framework Laptop 13 and Framework Laptop 16. The photo below highlights the difference in size between the two laptops.

The specification of my Framework Laptop 13 is outlined below, which offers exceptional performance and compatibility running Linux (NixOS), comparable to the Framework Laptop 16 across productivity, collaboration and web development tasks.

Framework Laptop 13 DIY Edition
AMD Ryzen 7 7840U (8C/16T)
64GB DDR5-5600 RAM
4TB Western Digital Black SN850X NVMe (7.3GB/s Read)
13.5-inch IPS LCD Display (2256x1504 @ 60Hz)
2x USB-C, 1x HDMI, 1x 1TB Expansion Card (1GB/s Read)

The only “disappointment” with this switch is the loss of the higher quality display, specifically the 165Hz refresh rate, which I would love to see as an option for the Framework Laptop 13.

Jensen Huang

Fri, 01 Mar 2024 00:00:00 -0600

In 2018, I completed my executive education at Stanford University. During my time on campus, I was fortunate to connect and engage with an incredible community, including my course peers and the world-class faculty.

Earlier today, the Stanford team released an interview (56 mins) with Jensen Huang, Founder and CEO of NVIDIA. With the current mainstream media hype regarding the use of Artificial Intelligence (AI), it is fascinating to hear NVIDIA’s journey, as well as Jensen’s insights into strategy and leadership.

NVIDIA has seen incredible success, driven by its tenacious attitude towards continuous innovation (identifying and solving big problems).

Outlined below are a few of my notes taken from the interview.

Core Beliefs: Establish, embed and maintain core beliefs, targeting a purpose that is meaningful with a clear/unique impact statement. Within that purpose, identify interesting problems that have the potential to become interesting markets. Solving these problems will unlock value and business growth.

Early Indicators of Future Success: Key Performance Indicators (KPIs) are difficult to define and understand, and commonly confused with actual results (e.g., Gross Margin is a result, not a KPI). Instead, businesses should identify Early Indicators of Future Success (EIoFS), which help provide finite milestones on an infinite journey, providing early evidence that the problem statement is solvable, leading towards a market opportunity.

First Principles: When looking to solve a problem, always go back to first principles (foundational understanding), accepting that the previously understood preconceptions may no longer be valid, superseded by new understanding and/or new technology.

Leadership: The role of a leader is to serve their team, and create the conditions by which others can do their best work. Trust is critical, respecting that each individual can accept and process complex information (good or bad). When successful, individuals will choose to work, without the need for exaggerated incentives.

Culture: In a traditional business, knowledge is power, which promotes a culture of secrecy. True power should be derived from an individual’s ability to reason complicated topics and empower others, leading them to greatness. Promote curiosity, collaboration and constructive reasoning. No task is beneath anyone, regardless of title and/or level.

Organisational Structure: Everyone within a business should look to maximise their personal value contribution, acknowledging that regardless of the title and/or level, everyone has a key role to play. Ignore other organisational structures, and focus on “Core Beliefs” and “First Principles” to define a structure that supports the unique business circumstance (NVIDIA is a very flat organisation and Jenson has fifty direct reports).

Innovation: Worry less about the past, always look towards the future and work backwards. Leverage “Early Indicators of Future Success” to measure progress and establish the required confidence to remain committed.

In conclusion, Jenson expressed the importance of having a clear purpose, with the goal of making a unique contribution.

This sentiment resonates strongly with me, an area that I have reflected upon over the years and recently written about in the article “Purpose and Fulfilment”.

ITPro Interview

Thu, 29 Feb 2024 00:00:00 -0600

In 2021, I collaborated with Boomi via a series of public engagements to discuss our business data integration story.

A summary of the published content can be found linked below.

Earlier this week, ITPro published a short interview follow-up.

How Elanco used newfound independence to rebuild its tech from the ground up

It is always fun to engage the community to share learnings, etc.

Framework Laptop 16

Sat, 03 Feb 2024 00:00:00 -0600

This week, the highly anticipated Framework Laptop 16 started shipping to “batch one” customers.

Thanks to our unique enterprise business partnership with Framework, Nirav and the team were kind enough to offer me early access. Therefore, over the past couple of weeks, I have been secretly testing a pre-release (near-final engineering sample) of the Framework Laptop 16.

I am not a reviewer, therefore if you are interested in a deep dive regarding the Framework Laptop 16, I recommend you watch the “Adam Savage’s Tested” overview video.

With that said, knowing the limited availability of the product, I thought I would share my initial thoughts, hopefully helping others considering a purchase.

I have split my thoughts under the headings “Positives” and “Constructive”. These thoughts have also been shared with Nirav directly.

Positives

Environmental Impact: Promoting the mission of the company, the Framework Laptop 16 proudly prioritises sustainability and the right to repair. Similar to the Framework Laptop 13, the care and attention demonstrated in the product design and manufacturing continue to impress, with thoughtful features that promote longevity through incremental upgrades and servicing. This is all achieved with minimal compromise regarding the performance, aesthetics and cost.

Expansion Cards: The increase from four expansion card slots on the Framework Laptop 13 to six on the Framework Laptop 16, offers unparalleled flexibility. With enterprise business usage in mind, it is common for I/O ports to suffer from “wear and tear”. Therefore, the option to easily replace them at a low cost is an added benefit. In addition, the new expansion card locking mechanism on the Framework Laptop 16 is more robust and user-friendly than the Framework Laptop 13.

Programmable Input Modules: The new programmable input module architecture has a lot of potential, building upon the established QMK firmware. Due to the pre-release nature of the product, the current real-world value is limited but could become an interesting differentiator for the Framework Laptop 16, assuming it is embraced by the community.

Input Module Connections: The input cover modularity is incredible, specifically the intuitive use of magnets and pogo pins. As a result, anyone can quickly and easily swap an input module or reconfigure the input layout with minimal effort, even while the laptop is in use.

Expansion Bay: The expansion bay is both brilliant and a little frustrating. Previous modular graphics solutions (used by Dell/Alienware) were always limited by the fact that the cooling solution was not part of the module itself, significantly limiting future upgrade options. Framework has designed a system that incorporates the graphics hardware (e.g., GPU/Memory) and the required active cooling system. This is brilliant, as it allows future graphics modules to “right-size” their cooling solution. As a negative, it means that the Framework Laptop 16 is not usable without a module installed. Therefore, if you plan/hope to “hot-swap” graphics whilst travelling, you will need purchase/carry an Expansion Bay Shell, making the total package a little cumbersome.

Native Linux Support: Framework continues to offer excellent support across Windows and Linux. Native Linux support is not always guaranteed by hardware manufacturers, resulting in conflicts and/or complex workarounds to enable network, audio and graphics. I can confirm the Framework Laptop 16 has excellent “out of the box” Linux compatibility.

PD 3.1 GaN Charger: Workstations and gaming laptops commonly have high power consumption requirements, resulting in proprietary connectors and large/heavy power adapters. The use of USB-C, as well as support for PD 3.1 and GaN are welcome additions to the Framework Laptop 16, delivering a compact 180W power adapter.

Performance: As my Framework Laptop 16 is pre-release (using a pre-release BIOS and drivers), I did not scrutinise the performance and/or thermal characteristics. With that said, I did complete a series of tests covering the CPU, GPU, Memory and Storage, which I can confirm all performed within the expected parameters. As the product matures, I expect these numbers to continue to improve.

Constructive

Size/Weight: Although the dimensions of the Framework Laptop 16 have been known for some time. The “real-world” experience of using the laptop, especially with the Graphics Module installed, emphasises the depth and weight. The larger dimensions are not uncommon with “gaming/workstation” laptops, but in recent years, the “premium” laptops do appear to be getting more compact and lighter. The photo below highlights the Framework Laptop 16 under an Apple MacBook Pro 16-inch (M2 Max). I do not see the size/weight as a “dealbreaker”, but certainly requires some adjustment (be prepared to get a larger carry case, especially when using the Graphics Module).

Keyboard Flex/Deflection: In use, my Framework Laptop 16 did exhibit some flex across the breadth of the keyboard, most noticeable at the edges closest to the input modules. Upon inspection, the mid-plate, which is used to mount the keyboard, is not particularly rigid (photo below). This presents a potential area of improvement, with the introduction of a more sturdy mid-plate.

Input Modules Wobble: When using the spacers (either side of the keyboard), there is a noticeable wobble (several millimetres of lateral movement). This does not impact the use of the laptop or the security of the input modules, but it does reduce the perceived “quality” of the build. I suspect this might be an engineering tolerance, that can be improved over time. In addition, knowing that many users will likely default to the central keyboard/trackpad configuration, I feel a full-width keyboard and trackpad input cover could be well received (removing any risk of wobble and improving the laptop aesthetics).

Expansion Bay Interposer: The custom expansion bay interposer is impressive engineering, unlocking the potential of PCIe, delivered in a package that can be easily removed with minimal risk of damage. With that said, transporting the interposer when not in use can be a delicate process (as each module requires a different interposer). I have found myself either screwing the interposer back into the unused Graphics Module or holding it until I find a safe location for storage. It would be great to have a small storage box for the interposer or a 3D printable model that could be used by the community.

Webcam, Microphone and Speaker Quality: I would describe the webcam, microphone and speaker quality as “good enough”, but not great. This is likely less important to gamers, who I suspect will use external peripherals. However, within the business environment, it is not uncommon for people to join meetings from “single-person” rooms, leveraging their integrated capabilities. Where possible, it would be great to see “premium” options made available on the marketplace. From a corporate standpoint, with the shift to FIDO2, I would also like to see an option for a Windows Hello compatible webcam. However, I appreciate this would be a Windows-specific component.

Conclusion

Overall, I have been impressed with the Framework Laptop 16 and commend the Framework team on their design/manufacturing expertise and dedication to sustainability.

Interestingly, my areas of critique are all viable for post-release continuous improvement. For example, when the Framework Laptop 13 first launched, I highlighted several points of frustration (e.g., glossy display, screen wobble, speaker volume, etc.) Over the coming months, Framework released new components that resolved each of these points, which not only demonstrates their customer-centric approach but also the unique value of a product designed with serviceability in mind.

As a result, I fully expect to see post-release improvements delivered for the Framework Laptop 16, which could be installed easily at minimal cost. I am not aware of any other laptop currently available on the market with this value proposition.

The only “fixed” restriction is the size/weight, acknowledging that the Framework Laptop 16 is large, especially when compared to an Apple MacBook Pro, Dell XPS or equivalent.

Congratulations again to the Framework team for the successful launch of the Framework Laptop 16! We received our first production batch for business use this week, which I am excited to get into the hands of users.

Thanks to Nirav, Chris and the entire Framework team for your continued partnership, helping us lead the charge to shift enterprise businesses towards sustainable computing.

Apple Vision Pro Impressions

Sat, 03 Feb 2024 00:00:00 -0600

In June 2023, I wrote an article reacting to the Apple Vision Pro announcement.

Although the Apple Vision Pro is not yet available in the UK, I am fortunate to have early access. This article will share my initial impressions.

If you are looking for an in-depth review, Nilay Patel from The Verge has posted an excellent video.

When first announced, Tim Cook (Apple CEO) described the Apple Vision Pro as “the first computer you look through, not at”.

I surmised that this statement accurately represented the ambition, but not the current reality. Having now used the Apple Vision Pro across a range of use cases, I can confirm my original assessment was accurate.

Outlined below are my thoughts…

Product Positioning

Over the past decade, I have used/owned a wide range of VR/AR headsets, including the latest headsets from Meta (Quest Pro and Quest 3) and HoloLens 2 from Microsoft.

Since the announcement, Apple has been eager to differentiate the Apple Vision Pro from other VR/AR headsets by calling it “Spatial Computing”. They have even banned developers from using the terms Virtual Reality and Augmented Reality within their application descriptions.

However, let us be clear, the Apple Vision Pro is 100% a VR/AR headset, which directly competes against other VR/AR headsets, most notably the Meta Quest range, etc.

Arguably, the Apple Vision Pro includes just two unique differentiators, specifically the front-facing lenticular display and the Solo Knit Band.

Excluding these features, when compared to the Meta Quest 3, the Apple Vision Pro offers a very familiar design and feature set.

With that said, I can confirm the Apple Vision Pro is the best, highest-quality VR/AR headset I have ever used. For example, the Sony Micro‑OLED displays are exceptional, with an astonishing 3,386 PPI (34 PPD). The pass-through cameras are also best-in-class (although not perfect), achieving an estimated 12ms response time.

However, this impressive specification is ultimately an evolutionary step, not revolutionary, as the form factor and usage of the headset are consistent with other VR/AR headsets.

It is also impossible to ignore the price, which starts at £3499 (256GB), which is 7x more than the Meta Quest 3 (128GB). Therefore, it is safe to assume that a Meta Quest 4 (or Meta Quest Pro 2) could close the specification gap, whilst continuing to be significantly cheaper.

Hardware

The hardware is beautiful, with exceptional build quality. The use of metal and glass makes the product look and feel premium, especially when compared to other VR/AR headsets on the market, which tend to favour plastic.

Unfortunately, although these premium materials look great, they hinder the use of the product itself. The Apple Vision Pro is undeniably heavy and frustratingly front-loaded, placing a lot of weight on the bridge of the nose and cheeks, which can quickly become uncomfortable during prolonged usage (> 30 minutes).

This issue is compounded by the Solo Knit Band, which looks great in promotional videos/photos (straight out of Ready Player One), but is less practical in use as it does not effectively distribute the weight of the headset. Apple were aware of this issue, as they include a less glamorous (but more practical) Dual Loop Band.

In addition, the front-facing lenticular display, combined with a software feature known as “EyeSight”, is an interesting concept, but offers very little real-world value. In theory, EyeSight reveals your eyes while wearing Apple Vision Pro, letting others know when you are using apps or fully immersed. Unfortunately, the display is difficult to see (especially in direct light) and the computer-generated eyes, although impressive engineering, are a little disconcerting.

Acknowledging these failings, it is ironic that the front-facing lenticular display and Solo Knit Band are what differentiates the Apple Vision Pro from other VR/AR headsets. I would not be surprised if they are both heavily revised (or even removed) in future generations.

Software

Unsurprisingly, at launch, the Apple Vision Pro software ecosystem is sparse, limited to a series of “tech demos” that fail to offer a compelling reason to use the headset for any meaningful length of time. The current range of apps are also very isolating, with no features that promote social engagement and/or collaboration.

A notable exception is Microsoft Office, which offers Apple Vision Pro optimised versions of Word, Excel, PowerPoint, Outlook, Teams, OneNote, and Loop. These apps work surprisingly well and I commend Microsoft for committing to the launch date.

Thankfully, the limited software ecosystem is likely an early adopter problem, which I am sure will improve quickly. I am however a little surprised Apple did not prioritise/promote a “killer app”, something that could showcase the capabilities of the Apple Vision Pro. The best example is probably media playback via Apple TV, which is very immersive, but not enough to differentiate from other VR/AR headsets.

Unlike other VR/AR headsets, the Apple Vision Pro does not include any physical controllers. Therefore, all interactions are via eye and hand tracking.

The initial setup process is more cumbersome than I would like, but once complete, everything works as advertised. For example, my eight-year-old son was able to navigate the user interface with minimal instruction.

Unfortunately, it is not perfect. Having never previously used my eyes as a direct human interface, I had not appreciated how frequently I interact with technology without directly looking at it. For example, thanks to muscle memory, I can tap, swipe and gesture on my iPhone/iPad, whilst my eyes focus elsewhere.

With Apple Vision Pro, this type of human interface does not work, as you must directly look at the user interface element when triggering an input.

As a result, although eye tracking does (at times) feel like “magic”, it can also be frustrating when attempting to complete more complex tasks. For example, when attempting to be productive whilst wearing the headset, I found I had to connect a physical keyboard/trackpad or interface with my Mac.

Overall, visionOS and the software ecosystem remind me of watchOS v1.0, which had plenty of promise but took several years to refine the experience.

Conclusion

The Apple Vision Pro is the most impressive VR/AR headset I have ever used! It is also the most expensive and uncomfortable, whilst currently offering very little real-world value.

As a result, in February 2024, it is impossible to recommend the Apple Vision Pro. If you are in the market for a VR/AR headset, I would point you towards the Meta Quest 3, which is a fraction of the cost and has a comprehensive software ecosystem.

However, that does not mean the Apple Vision Pro is a failure. I continue to see the promise, which I believe has been best represented in the recent Casey Neistat video.

There is a world where “Spatial Computing” becomes viable, delivering unique, differentiated experiences that augment traditional computing concepts.

Unfortunately, I believe we are at least four generations (8+ years) from that reality, where the physical limitations of the headset have been mitigated and the software experience has been refined and expanded.

The path from today towards that future is exciting but also expensive and fraught with missteps. Therefore, it will be interesting to watch Apple lead from the front vs. joining at a later date, once the market has a solid foundation.

Framework and NixOS

Thu, 01 Feb 2024 00:00:00 -0600

In the article “Linux Endgame - NixOS”, I outlined my rationale and initial impressions testing NixOS on my Framework Laptop 13 (AMD Ryzen).

As highlighted in the post, NixOS is a free and open-source, immutable Linux distribution based on the purely functional Nix package manager. NixOS is composed using modules and packages defined in the nixpkgs project.

As a long-time Fedora user, the switch to NixOS has been a little jarring. In many respects, I have had to “unlearn” many common Linux workflows, such as package management, upgrades, etc.

I would also argue that the documentation supporting NixOS is not yet equivalent to that offered by the Fedora and/or Arch communities. Therefore, overcoming inevitable challenges can require more effort and research.

In part, the lack of documentation is why I decided to write this article, which aims to outline my process of installing NixOS on a Framework Laptop 13, including my common software. At this time, I have not included more advance NixOS features such as Flakes and Home Manager. However, I do intent to write about these in the future.

The full specification of my Framework Laptop 13 can be found below. The steps outlined in this post should work for any AMD-based Framework Laptop (including the Framework Laptop 16).

Framework Laptop 13 DIY Edition
AMD Ryzen 7 7840U (8C/16T)
64GB DDR5-5600 RAM
4TB Western Digital Black SN850X NVMe (7.3GB/s Read)
13.5-inch LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

Before installation, it is important to disable Secure Boot from the Framework BIOS, otherwise the NixOS installation media will not be detected. This is achieved by tapping the “F2” key at boot (not the “F12” key), which will prompt the option “Administer Secure Boot”.

With Secure Boot disabled, the first step is to download and install NixOS. Thankfully, this task is very similar to other Linux distributions, facilitated by a “live CD”, which can be used to create a bootable USB drive.

The “live CD” runs the GNOME desktop environment and includes a full GUI installation wizard, whilst also allowing you to test NixOS before the actual installation (which can help identify any potential hardware conflicts, etc.) It should be noted, that any supported desktop environment (not just GNOME) can be selected during the installation process.

By default, I recommend full disk encryption on all modern systems. Unfortunately, unlike Fedora and Pop_OS!, the NixOS GUI installer does not include a “one-click” encryption option. As a result, full disk encryption must be configured manually.

I have outlined the nine-step process to enable full disk encryption in the article “NixOS Full Disk Encryption”.

With NixOS successfully installed and following a reboot, you will be greeted by a very clean/minimal operating system, with a limited number of packages (applications) installed. By default, you do not get common applications such as LibreOffice, etc.

The unique proposition offered by NixOS is that the entire operating system, including the kernel, applications, system packages, and configuration is declared in a configuration file, using the Nix language. The primary configuration file can be found at the following location and will be used to customise the setup per your requirements.

/etc/nixos/configuration.nix

At this stage, it is worth becoming familiar with the following terminal commands. Although fairly self-explanatory, I will describe the use of these commands throughout this post.

sudo nano /etc/nixos/configuration.nix

sudo nixos-rebuild switch

sudo nixos-rebuild boot --upgrade

sudo nixos-rebuild switch --rollback

sudo nix-collect-garbage -d

sudo nix-collect-garbage --delete-older-than 30d

flatpak update -y\

Edit the “configuration.nix” file using nano sudo nano /etc/nixos/configuration.nix and add the following configuration lines to install and enable key services, specifically Fingerprint Reader, Flatpak, Docker, virt-manager, and OneDrive.

As a reference, I have made my “configuration.nix” file available on GitHub.

services.fprintd.enable = true;  

services.flatpak.enable = true;  

virtualisation.docker.enable = true;  

virtualisation.libvirtd.enable = true;  
programs.virt-manager.enable = true;  

services.onedrive.enable = true;  

The following configuration installs and enables custom fonts, including the common Microsoft fonts (useful when viewing/editing Microsoft Office documents). Any new fonts can be added to this section.

fonts.packages = with pkgs; [  
  corefonts  
  vistafonts  
  noto-fonts  
  noto-fonts-cjk  
  noto-fonts-emoji  
  liberation_ttf  
  fira-code  
  fira-code-symbols  
  mplus-outline-fonts.githubRelease  
  dina-font  
  proggyfonts  
];

fonts.fontDir.enable = true;

The following configuration installs and enables packages from NixOS Packages. Where possible, I prioritise Flatpak applications, installed via Flathub. Therefore, I only use the Nix Package Manager when an application is not available and/or viable (e.g., feature-complete, reliable or performant) as a Flatpak.

environment.systemPackages = with pkgs; [  
  wget  
  git  
  google-chrome  
  vscode  
  ollama  
];

Flatpak applications do not need to be declared within the “configuration.nix” file, as they are not part of the base image.

As a result, Flatpak applications are installed and maintained from the command line and/or GNOME Software store, similar to other GNOME-based Linux distributions. Once installed, Flatpak application icons may not appear until after a reboot. They will also need to be manually updated using the ‘flatpak update -y' command.

A full list of the Flatpak applications I use can be found in the article “Flatpak Availability”. However, for reference, the commands to install my most common applications for NixOS can be found below.

flatpak install flathub org.libreoffice.LibreOffice  
flatpak install flathub com.github.IsmaelMartinez.teams_for_linux  
flatpak install flathub com.slack.Slack  
flatpak install flathub org.mozilla.firefox  
flatpak install flathub md.obsidian.Obsidian  
flatpak install flathub org.standardnotes.standardnotes   
flatpak install flathub org.gimp.GIMP  
flatpak install flathub org.kde.krita  
flatpak install flathub org.audacityteam.Audacity  
flatpak install flathub org.videolan.VLC  
flatpak install flathub fr.handbrake.ghb  
flatpak install flathub org.openshot.OpenShot  
flatpak install flathub fr.natron.Natron  
flatpak install flathub com.obsproject.Studio  
flatpak install flathub org.gnome.Boxes  
flatpak install flathub org.filezillaproject.Filezilla  
flatpak install flathub com.getpostman.Postman  
flatpak install flathub com.heroicgameslauncher.hgl  

With the “configuration.nix” file updated, exit and save the file by pressing ‘^ + X’ (if using Nano) and run the command sudo nixos-rebuild switch. This will verify the “configuration.nix” and build/switch to the new configuration. If an issue is identified within the “configuration.nix” file, an error message will be displayed.

Finally, with the newly defined configuration operational, run the following terminal commands to finish the setup of specific services/packages. Specifically, they enable the Flathub repository, allow the custom fonts to be accessed by Flatpak applications and set up OneDrive with your Personal or Business account.

flatpak remote-add --if-not-exists flathub https://dl.flathub.org/repo/flathub.flatpakrepo

ln -s /run/current-system/sw/share/X11/fonts ~/.local/share/fonts  
flatpak --user override --filesystem=$HOME/.local/share/fonts  
flatpak --user override --filesystem=$HOME/.icons

onedrive  
systemctl --user enable onedrive@onedrive.service  
systemctl --user start onedrive@onedrive.service  
systemctl --user status onedrive@onedrive.service

That’s it! As a final step, it is worth running the command ‘sudo nixos-rebuild boot –upgrade’, which will install any system updates and reboot. It should also “tidy” any loose ends as part of the step, specifically the enablement of Flatpak application icons, etc.

The final point for reference is the unique rollback option provided by NixOS. Thanks to the immutable nature of the operating system, every time a new configuration is switched to running (using the command ‘sudo nixos-rebuild switch’), the previous configuration is saved.

When rebooting, every previous configuration state is made available. Therefore, if you ever encounter an issue, the process to roll back is as simple as selecting a previous working configuration.

This is incredibly powerful, but can also lead to a lot of old configurations being stored. Therefore, the terminal command ‘sudo nix-collect-garbage -d’ will delete all unused configurations. You can also specify certain criteria, for example, ‘– delete-older-than 30d’, which will delete anything older than 30 days.

In theory, NixOS should be fully operational on the Framework Laptop 13, with all hardware (e.g., Fingerprint Reader, Webcam), services and applications enabled!

Flatpak Availability

Wed, 31 Jan 2024 00:00:00 -0600

Earlier this month, I shared an update regarding my common software list.

The list includes multiple operating systems, including Linux. I am a long-time Fedora user, but am currently exploring the world of NixOS.

Regardless of the specific Linux distribution, I have always been an advocate for application containerisation, specifically Flatpak.

The list below outlines which applications from my software list are currently available as a Flatpak from Flathub.

A verified Flatpak means that it was published and is maintained by the original developer or a third party approved by the developer (making it official).

Any Flatpak that is not verified does not make it “bad”, but means a third party (unaffiliated with the original developer) has published and is maintaining the application as a Flatpak.

Flathub does review newly published Flatpak applications (providing some assurance), but I would also ensure any non-verified Flatpak has a healthy change log (recent updates) and usage (statistics).

To help with the installation of these applications, I have added the relevant commands below.

flatpak install flathub org.libreoffice.LibreOffice  
flatpak install flathub com.github.IsmaelMartinez.teams_for_linux  
flatpak install flathub com.slack.Slack  
flatpak install flathub com.visualstudio.code  
flatpak install flathub com.google.Chrome  
flatpak install flathub org.mozilla.firefox  
flatpak install flathub md.obsidian.Obsidian  
flatpak install flathub org.standardnotes.standardnotes  
flatpak install flathub com.bitwarden.desktop  
flatpak install flathub org.gimp.GIMP  
flatpak install flathub org.kde.krita  
flatpak install flathub org.audacityteam.Audacity  
flatpak install flathub org.videolan.VLC  
flatpak install flathub fr.handbrake.ghb  
flatpak install flathub org.openshot.OpenShot  
flatpak install flathub fr.natron.Natron  
flatpak install flathub com.obsproject.Studio  
flatpak install flathub org.gnome.Boxes  
flatpak install flathub org.filezillaproject.Filezilla  
flatpak install flathub com.getpostman.Postman  
flatpak install flathub org.godotengine.Godot  
flatpak install flathub com.heroicgameslauncher.hgl  
flatpak install flathub com.usebottles.bottles  
flatpak install flathub com.github.tchx84.Flatseal

Any application that is not available as a Flatpak must be installed via another package manager (e.g., DNF, Nix Package Manager, etc.)

NixOS Full Disk Encryption

Tue, 30 Jan 2024 00:00:00 -0600

Yesterday, I shared a short article regarding NixOS.

Over the next couple of weeks, I plan to share my experience using the operating system, specifically with my Framework Laptop 13 (AMD Ryzen).

Unlike Fedora and Pop_OS!, the NixOS GUI installer does not include a default option for full disk encryption. Therefore, similar to the process I outlined for Manjaro, the encrypted partitions must be manually created.

Thankfully, this is a relatively easy process, as outlined below.

Download the NixOS Graphical ISO image, specifically the GNOME 64-bit Intel/AMD. NOTE: This does not force the use of GNOME. Any supported desktop environment can be selected as part of the installation.
Create a bootable USB.
Ensure Secure Boot is disabled on the target system and boot from the USB.
Connect NixOS to the Internet (Wired or Wireless). Restart the NixOS installer application if the Internet connection is not automatically detected.
Proceed through the installation wizard, until partitioning.
Select “Manual Partitioning” and click “Next”.
Delete any existing partitions from the target drive. NOTE: All previous data will be lost.
Click “Create Partition Table” and select “GUID Partition Table (GPT)”.
Click “Create” to create a new partition using the following configuration. Repeat this step until all three partitions have been created.

Partition One: Boot

Size: 512 MiB
Partition Type: GPT
File System: fat32
Encrypt: Not Checked
Mount Point: /boot
Flags: boot

Partition Two: Swap

Size: 8192 MiB (Minimum)
Partition Type: GPT
File System: linuxswap
Encrypt: Checked (Enter Username/Password)
Mount Point:
Flags: swap

Partition Three: Root

Size: Remaining Drive Space
Partition Type: GPT
File System: btrfs
Encrypt: Checked (Enter Username/Password)
Mount Point: /
Flags:

That’s it! Click “Next” and complete the installation process. You may receive a warning regarding the non-encrypted boot partition, this can be ignored.

At boot, the system will prompt for the encryption password, before launching NixOS (where you will be prompted for your operating system password).

Linux Endgame - NixOS

Mon, 29 Jan 2024 00:00:00 -0600

I began my Linux journey in 2001 with Mandrake Linux 8.1. Over the past 23 years, I have “distro hopped” a fair amount, covering distributions based on Red Hat, Debian and Arch.

With my background in enterprise business, I always found myself returning to Fedora as my daily driver, mainly driven by my familiarity with the Red Hat Enterprise Linux (and the DNF package manager), as well as the fact that I am a long-time GNOME user (starting with GNOME 1.4.1).

In 2021, I flirted with Fedora Silverblue, which is an immutable operating system from Red Hat (IBM).

The purpose of an immutable operating is to ensure the host is protected from accidental or malicious modifications, improving security, stability and performance by guaranteeing the integrity of the core system.

I am very supportive of this approach and am not surprised to see it become a “standard” with modern operating systems.

Unfortunately, I could never make Fedora Silverblue operate well with my workflow. For example, Fedora Silverblue is essentially one big system image, which expects all applications to be installed in a container via Flatpak. This is a great architecture, except for when an application/service is not available (or viable) as a Flatpak. For example, Node.js, Docker, Virt-manager, etc.

In this scenario, you have two primary options. Use rpm-ostree to add a new layer to the base image with the required packages or create a mutable environment running on Fedora Silverblue. I found rpm-ostree overly complicated to manage and the process of creating “walled gardens” for specific applications/services has an impact on security and usability.

Over the past 18 months, the LINUX Unplugged community has become increasingly passionate about NixOS, it even won their Linux distribution of the year at the 2023 Tuxies.

I have listened intently, but only recently had the available time to test NixOS. I also caution anyone within the Linux community about “jumping on the bandwagon” too early, as it takes time for a Linux distribution to prove itself as a viable long-term option.

NixOS is a free and open-source, immutable Linux distribution based on the purely functional Nix package manager. NixOS is composed using modules and packages defined in the nixpkgs project.

Similar to Fedora Silverblue, the NixOS base image is immutable and read-only. However, the individual user can decide what is included within that immutable image by following a declarative configuration model. For example, the entire operating system, including the kernel, applications, system packages, and configuration is declared in a configuration file (typically located in /etc/nixos), using the Nix language.

The advantage of this approach is that it creates a reproducible system configuration, which can be maintained and versioned as code. This configuration improves visibility, consistency and integrity of the running configuration, whilst also providing a recovery and migration path that is guaranteed and very easy to trigger.

Similar to Fedora Silverblue, NixOS still promotes the use of Flatpak, providing a clear separation between the base image and applications. However, with 80,000+ packages available as part of the Nix package manager, NixOS is incredibly flexible and versatile.

I have been testing NixOS on my Framework Laptop 13 (AMD Ryzen) and have been incredibly impressed. I have created a fully-encrypted, lightweight/minimal Linux distribution (running GNOME) that feels just like Fedora, but with the added benefit of being immutable, whilst still incorporating every application/service required for my workflow.

In addition, the configuration I have established is maintained as code within GitHub, therefore I can easily modify and experiment, knowing that this configuration can be deployed to any number of systems.

Over the coming weeks, I plan to share some of my initial findings, including my current configuration. Stay tuned!

Software List 2024

Sat, 20 Jan 2024 00:00:00 -0600

In 2020, I shared a list of my most commonly used software, across macOS, Windows, Linux, and iPadOS.

Although I have updated the list a couple of times over the past few years, I thought it was time to complete a full refresh.

As highlighted, I still use all the major desktop operating systems regularly (specifically macOS, Windows 11 and Fedora). As a result, where possible, I target free and open-source software (FOSS) that is available cross-platform. This allows me to have a consistent experience when switching between operating systems.

Unfortunately, there is not always a viable FOSS option or a specific platform may have native software that offers unique value. For example, Xcode, which is only available on macOS.

I feel very comfortable recommending all the software on this list as I have used the majority for many years.

Framework and Windows To Go

Mon, 15 Jan 2024 00:00:00 -0600

I am the proud owner of a Framework Laptop, running Linux (Fedora). I have written a range of articles about Framework (listed below), covering my personal use and newly established enterprise business partnership.

Framework and Linux offer an incredible combination of open hardware and software that works exceptionally well together.

However, occasionally I still need access to Windows, usually to run a specific software or to test a configuration. Virtualisation and/or a translation layer such as Wine are viable options, but each includes specific limitations that can be cumbersome/frustrating to manage.

Another option would be to dual-boot Linux and Windows. However, the two operating systems partitioned to run on a single storage device can cause problems, especially during operating system upgrades, etc. As a system that I rely upon, having to deal with boot conflicts and/or data integrity issues is something I aim to avoid.

Thankfully, I have found an alternative that works incredibly well, a combination of Windows To Go and the official Framework Storage Expansion Card.

Windows To Go is a forgotten feature of Windows 10/11 that allows Windows to be installed and used from a USB-connected external drive (removing the need for internal storage). The Framework Storage Expansion Card, specifically the 1TB capacity, is the perfect companion, as it includes a custom controller that delivers 1GB/s read and write speeds.

This is not as fast as a modern-day NVMe SSD but offers more than enough performance to run Windows, including I/O-intensive applications.

For example, the benchmark results below are from my Framework Laptop 13 running Windows To Go (Windows 11) from a 1TB Framework Storage Expansion Card.

Not only is Windows performant, but it is also surprisingly reliable. As an extreme example, it is possible to physically remove the Framework Storage Expansion Card whilst Windows is running. The active session of Windows will “stall”. However, upon reinsertion of the Framework Storage Expansion Card, the Windows session immediately resumed (I was amazed this worked).

The installation of Windows To Go on the Framework Storage Expansion Card is very simple. However, you will need a Windows PC for the setup.

Download Rufus for Windows and the latest Windows 11 ISO from Microsoft.
Insert the Framework Storage Expansion Card into the Windows PC and launch Rufus.
Select the Framework Storage Expansion Card as the “Device” and the Windows 11 ISO as the “Boot Selection”. If the Framework Storage Expansion Card does not appear in the device list, select “Show Advanced Driver Properties” and check “List USB Hard Drives”.
Under “Image Option”, select “Windows To Go”. All other options can be left default.
The Framework Storage Expansion Card can now be safely removed and inserted into the Framework Laptop.

There is no need to modify the Framework Laptop bootloader configuration or any BIOS settings, simply hold F12 at POST and select the Framework Storage Expansion Card.

The great thing about this approach is that Linux and Windows never compete for resources or risk any configuration conflicts. If you ever want to purge Windows, simply remove the Framework Storage Expansion Card.

In addition, the Framework Storage Expansion Card can be swapped into other Framework Laptops (or any device with a USB 3.3 Gen 2 interface). Windows will need to re-configure upon launch, but everything should work as designed.

Regarding limitations, there are very few. Microsoft has published a comprehensive Windows To Go FAQ. The main consideration is BitLocker encryption, which can be enabled, but does not leverage secure boot or the TPM. This limitation is worth acknowledging, but the default software encryption is likely strong enough for the majority of users.

All other Windows features work as designed, including Windows Update (allowing you to run the latest Windows builds, etc.)

Overall, I have found Windows To Go installed on the Framework Storage Expansion Card to be a great experience. Highlighting the versatility of the Framework Laptop and its modular architecture.

balenaEtcher Recovery

Fri, 12 Jan 2024 00:00:00 -0600

As someone who tests a lot of different devices (desktops, laptops, single-board computers) and operating systems (Windows, macOS, Linux), I create a lot of bootable USB drives.

Although I believe Rufus for Windows is a superior tool for creating bootable images, I commonly use balenaEtcher as it supports Windows, macOS and Linux.

The tool itself works great and can be used for a wide range of purposes. For example, creating bootable Linux USB drives and/or flashing operating systems to be used with single-board computers (e.g., Raspberry Pi , etc.)

However, the resulting bootable USB is often unreadable when looking to reformat for other purposes.

Thankfully, when this scenario occurs, it is easily resolved using the following commands.

Windows

Open a CMD prompt and run the command diskpart.exe.
Run the command list disk.
Run the command select disk x, where “x” is the disk ID shown in the previously generated list.
Run the command clean, which will erase the file system.

macOS

Open Terminal and run the command diskutil list.
Run the command diskutil eraseDisk FAT32 UNTITLED MBRFormat /dev/diskX, where “diskX” is the target USB drive.

Linux

Open Terminal and run the command sudo dd if=/dev/zero of=/dev/xyz bs=512 count=1 conv=notrunc, where “abc” is the target USB drive device path.

Upon completion of these steps, the USB drive should respond as normal, accessible to any operating system for reformatting.

Arc for Business

Wed, 10 Jan 2024 00:00:00 -0600

In March 2023, I wrote a short article regarding my usage of Arc, a Chromium-based browser from The Browser Company.

Nearly one year later, I continue to use Arc as my primary browser on macOS, displacing Google Chrome. In addition, I have been working with The Browser Company to bring Arc to business, with early access to the Windows beta.

I am pleased to report that Arc continues to deliver on its promise, taking the best parts of Google Chrome, whilst restructuring the user experience to better suit modern web workflows.

As highlighted in my previous article, my greatest concern for the future of Arc is the pressure to generate money for investors. This can lead to shareholder-centric features that offer minimal end-user value, whilst also impacting the core experience.

In my opinion, this is exactly what happened with Microsoft Edge, which had the potential to be a great Chromium-based browser, but has been compromised by the “internal politics” at Microsoft, which prioritised adjacent agendas (e.g., advertisement revenue, co-promotions, referrals, etc.)

In October, I feared my concerns had been realised when The Browser Company released Arc Max, which is a collection of integrated Generative AI features. I completely appreciate the “hype” associated with Generative AI, resulting in widespread FOMO. However, in my opinion, prioritising “non-essential” features based on hype alone is a slippery slope, that can quickly become a distraction, leading to “bloatware” that increases the complexity of the codebase, ultimately impacting performance, reliability and security.

Thankfully, Arc Max was implemented fairly well, remaining optional (off by default), with a non-intrusive user interface and no immediate requirement to subscribe and/or pay for use. I have chosen to be optimistic, hoping that this implementation approach is evidence that The Browser Company understand the importance of protecting their core value proposition (user experience).

As a comparison, Arc for Windows (which is still in beta) is not at feature parity with the macOS version, highlighting the potential of a browser that is streamlined, lightweight and focused.

In my opinion, The Browser Company should take inspiration from Arc for Windows and not aim for feature parity with the macOS version. Instead, I would like to see the two versions “meet in the middle”, looking for opportunities to remove/consolidate features from Arc for macOS.

Unfortunately, I fear the pressure to generate money for investors will continue to build. At this time, it would appear The Browser Company are hoping that scale can help overcome this challenge, which commonly unlocks new monetisation opportunities (search deals, referrals, etc.) However, as proven by Mozilla and Firefox, I do not believe scale alone is the answer to sustained success, as it can be easily lost. For example, in 2010, Firefox had an estimated 32% market share, only to see that reduce to 3% in 2024.

Therefore, I would like to see The Browser Company supplement Arc development with a sustainable business model, whilst protecting against the pressure to develop/implement features that are shareholder-centric (self-serving, not customer-serving).

This is where I believe an enterprise business opportunity has potential. For example, the majority of Fortune 500 enterprise businesses have a heavy reliance on Software-as-a-Service (e.g., Office 365, ServiceNow, Workday, Salesforce.com), which are commonly accessed via the browser. Arguably, the browser is the most important enterprise business application, maybe more important than the underlying operating system.

Therefore, if specific features can be developed that target security, compliance, and centralised management, I feel there is the potential for Arc to become the “default” browser for enterprise businesses, reducing the reliance on Google and Microsoft.

These features would not be “core” to Arc, but instead offered as modules and/or optional add-ons. Therefore they would not impact the consumer experience and/or the integrity of the foundational codebase. They could also be developed by a dedicated business development team, delivering a clear separation of concerns.

Specific enterprise business features could include:

Centralised Management (Configuration Toggles, Sidebar/Spaces Customisation, Extensions Management, etc.)
Specific Security Controls (Trusted Sites, Force HTTPS, etc.)
Azure AD Integration with FIDO2 Compatibility (SSO Management, etc.)
Block Lists (Sites, Downloads, etc.)
Ability to Disable Arc Features (Easels, Notes, Boosts, Max, etc.)
Max Integration with a Private LLM (Azure OpenAI, Google Vertex AI, etc.)
Private Boost Store (Library of Enterprise Approved Boosts)
Custom Branding (Business Logo, Colours, Icons, etc.)
Cross Platform Support (Windows, macOS, Linux, iOS, iPadOS, Android, etc.)
Encrypted Sessions (Tunnelling) and Integration with SASE Solutions
Data Loss Prevention (DLP) Controls

These features would be compelling to most enterprise businesses, as they would improve security and compliance, whilst also enhancing the user experience and prompting agnostic endpoints (decoupling from the underlying operating system).

It is also a viable monetisation strategy, as we know enterprise businesses are accepting of pay-per-user subscriptions (assuming the right price point). With millions of users as the target audience, The Browser Company would only need to secure a small number of strategic partnerships to provide a robust, reoccurring revenue stream that should please investors.

This would allow the core development team to stay focused on customer-centric features that further enhance/enrich the product, hopefully resulting in greater scale, etc.

It should be noted that browsers with a focus on enterprise businesses do already exist (e.g., Talon). However, they are usually generic copies of Google Chrome, that fail to offer any additional user benefits. This is a major barrier, as enterprise business users are still users and ultimately desire products that enhance their productivity, etc. This is where Arc is different, as it has a unique and compelling value proposition regarding user experience.

Generative AI - EoY

Fri, 15 Dec 2023 00:00:00 -0600

Earlier this year, my team released a Generative AI chatbot for business, focused on productivity acceleration by streamlining common tasks. I documented the design and development process through a series of blog posts (links below).

As we close the year, I thought I would take a moment to reflect on our progress.

As a reminder, the Generative AI chatbot for business was designed and built by two engineers (working part-time) and released to our business (7000 viable users) in June 2023.

The high-level diagram below highlights the architecture, which is multi-cloud (Microsoft Azure and Google Cloud Platform) by design. It includes a responsive front-end and custom middleware, with each service being loosely coupled and ephemeral, maximising flexibility and cost efficiency.

The entire application is managed as Terraform, meaning it is cloud-agnostic and can be deployed “as-code”. The only exception is the language models, which are cloud-specific (e.g., Azure OpenAI Service, Google Vertex AI, etc.)

Regarding features, the list below outlines the current state.

Azure AD Authentication (SSO/FIDO2/MFA Compatible)
User Authorisation
Secure/Hardended Codebase and Environments (Azure and GCP)
Logging and Application Monitoring
Cost Monitoring
Responsive User Interface and Middleware API
Ephemeral Services
Infrastructure-as-Code (Hashicorp Terraform)
Multiple Models (GPT-3.5, GPT-4, GPT-4 32k, GPT-4-Turbo, PaLM2, Imagen 2)
Maximum 128K Token Limit
Multimodal (Text and Images)
User Context Caching
Obfuscated User Usage Data
Plain Text Import/Export
File Upload
Custom Personas
Custom Persona Creation Wizard
Custom Persona Marketplace
Context Ingestion (Business Data)
Dashboard (Usage, Return on Investment)

Note: PaLM2 is in the process of being migrated to the recently announced Google Gemini Pro and Ultra.

Highlights include the controls that ensure security, privacy, and compliance, alongside the custom persona capability, which allows anyone to create their own Generative AI chatbot, primed for a specific use case, with pre-defined context. This feature is similar to what OpenAI recently announced, called GPTs.

In addition, the decoupled architecture allows us to add/replace/modify large language models very easily. For example, GPT-4-Turbo was made available to all users the week of release. This flexibility extends to the custom persona capability, meaning that any new large language model can be instantly used with a custom persona without any engineering intervention.

The screenshot below highlights the basic user interface, which is fully responsive, offering quick links to standard and custom personas, as well as model selection. Certain personas will default to a specific model, attempting to ensure the best user experience.

As evident, the user interface is inspired by ChatGPT, helping to reduce any requirement for training, etc.

Earlier this month, we enhanced our custom persona capability, which allows anyone to create their own Generative AI chatbot. We have seen strong adoption of this feature, with over 500 custom personas created, covering a wide range of use cases. The screenshot below highlights the creation wizard.

To ensure custom personas are discoverable and accessible, we have also implemented a marketplace (screenshot below). The creator can decide to publish their custom persona to the company, whilst also having the option to edit.

As an example, I have created a custom persona that ingests our Information Security architecture documentation. Therefore, I can simply upload any public security alert (e.g., notification, bulletin, briefing) to get a quick assessment of the risk, specific to our business. Considering I receive thousands of security updates across multiple channels, this custom persona is a great time saver.

Finally, the screenshot below is the dashboard on 15 December 2023.

Since June, we have seen 3,877 unique users access the service (55% of the viable audience), with approximately 800 to 1000 logins per day. That includes 538 custom personas active and broad usage of GPT-3.5, GPT-4, GPT-4-Turbo and Imagen 2. It will be interesting to see the Gemini Pro and/or Ultra growth in 2024.

Based on our conservative estimates, this usage has driven over 10,000 hours of time-saving, which is the equivalent of $499,144 of savings through productivity enhancements and cost avoidance.

Considering this adoption has been organic (no formal organisational change management) and the service only costs approximately $120 per month to operate (all users), I am very impressed with the return on investment.

Overall, I am very pleased with the progress we have made in 2023. I am pleased we chose to “lean in” to Generative AI and I am excited to see what additional value we can drive in 2024!

XREAL Air 2 Pro

Fri, 01 Dec 2023 00:00:00 -0600

Over the years, I have experimented with different head-mounted displays, usually focused on Virtual Reality (VR) and/or Augmented Reality (AR).

Recently, I have been testing the XREAL Air 2 Pro wearable display, which delivers a large, high-resolution virtual display (like having a portable monitor). Unlike other VR/AR headsets, it does not include any integrated computing capabilities, therefore the XREAL Air 2 Pro must be connected (USB-C) to another device (e.g., Laptop, Smartphone, Games Console, etc.)

Without the need to integrate computing capabilities, the XREAL Air 2 Pro is a smaller device, that attempts to mimic the look and feel of standard sunglasses.

In reality, I would argue they are “oversized” sunglasses, certainly more discrete than a Meta Quest Pro, Microsoft HoloLens, or Apple Vision Pro, but still not something I would want to wear casually.

The specifications of the XREAL Air 2 Pro can be found below. The list price in the UK is £449. XREAL also offer the XREAL Air 2 at a list price of £399 in the UK.

Display: Sony 0.55 Micro-OLED Panel
Field of View: 46-degree
Virtual Screen Size: 330-inch
Refresh Rate: 120Hz
Brightness: 500 Nits
Audio Output: Second Generation Speakers
Audio Input: Two Microphones
Input: USB-C
Weight: 75g (50:50 Weight Distribution)

The primary difference between the XREAL Air 2 and XREAL Air 2 Pro, is that the “Pro” supports three electrochromic dimming levels (0%/35%/100%), helping to deliver the best possible experience across different lighting conditions (indoors or outdoors). With that said, I am not sure this feature is worth the additional £50.

The XREAL Air 2 Pro supports four display modes. However, only “Air Casting” is supported natively, with all other modes requiring the XREAL Beam accessory. Air Casting is limited to a 130-inch vertical screen size.

Air Casting: Highest refresh rate (120Hz), display instantly follows your head movements.
Anchor: Reduced refresh rate (72Hz), display is fixed.
Follow: Reduced refresh rate (72Hz), display smoothly follows your head movements.
Sideview: Reduced refresh rate (72Hz), small display (20-inch) off-centre.

The XREAL Beam accessory is sold separately (£119). It unlocks the additional display modes (Anchor, Follow and Sideview) and includes an integrated battery (rated for 3.5 hours of use), as well as 32GB of storage, which can be used to install and run specific Android apps (e.g., Netflix, etc.) At the time of writing the number of officially supported Android apps, is very limited.

The XREAL Beam is also required if you want to use the XREAL Air 2 Pro with a games console (Sony PlayStation 5, Microsoft Xbox Series X/S, Nintendo Switch, etc.) Therefore, depending on your use case, the XREAL Beam may be a necessity.

Today, I have a monitor at the office and home. In either scenario, one monitor is always under-utilised (unused on the days I am at the other location). It is possible to mitigate this “waste” with shared workspaces, but this requires a level of coordination that is not always viable. With an XREAL Air 2 Pro the monitor becomes “portable”, easily transported between locations (achieving 100% utilisation).

If we assume a list price of £250 for a single 27-inch 4K monitor (e.g., Dell S2721QS), that would £500 for two (assuming only one at each location). Therefore, the XREAL Air 2 Pro at a list price of £449 would deliver a positive return on investment.

When connected directly to my MacBook Pro or iPhone 15 Pro via USB-C, the Sony 0.55 Micro-OLED panel is beautiful, with great colours and image quality (sharp text, etc.) This is especially apparent when viewing video or gaming.

Unfortunately, when attempting productivity tasks, such as web browsing, word processing, and coding, the experience is more challenging. Although the glasses are adjustable, including the angle of the arms and different-sized nose pads, I found it very difficult to find the perfect position. When not positioned correctly, the edges of the display (top/bottom and corners) become obstructed (outside the field of view) or blurry.

This is less of an issue when viewing video or gaming, as these areas are rarely the focal point. However, when focused on productivity tasks, many operating system and/or application controls are positioned at the edge.

Therefore, although it is not unusable for productivity, I have found it to be a frustrating experience, impacting my willingness to persevere for prolonged periods.

I had hoped the XREAL Beam accessory might help, as it provides more fine-grain adjustment, as well as the ability to simulate multiple monitors (which is cool). However, when using the Bean, the refresh rate is immediately halved (60Hz) and the image quality appears reduced (colour accuracy, etc.) This is disappointing, as the previously mentioned benefits are sacrificed for the increased flexibility.

In addition, the field of view (46-degree) remains fixed. Therefore, when adjusting the size and/or position of the display, the output is more prone to clipping (where the display moves outside of the viewable field of view.

In short, although the XREAL Beam has potential, I feel the negatives introduced are currently a deal breaker. Maybe these frustrations can be improved over time through software updates.

In conclusion, I do not consider the XREAL Air 2 Pro as a viable display for productivity use, leaving just video playback and gaming. This makes the list price of £449 less desirable and not something I can recommend. As for the XREAL Beam, at this time, I do not believe it is worth the additional £119.

Although my experience has been disappointing, I still believe the “concept” of a wearable display is interesting and worth continued exploration. Therefore, I hope XREAL and others continue to invest and innovate.

For example, I can imagine a world (likely many years from now) where VR/AR headsets and wearable displays such as the XREAL Air 2 Pro converge. In this scenario, the headset would target the form factor or the XREAL Air 2 Pro and/or Meta Smart Glasses, with more robust VR/AR capabilities. I can only assume this is what Apple is targeting with the soon-to-be-released Apple Vision Pro.

What I Do

Sat, 25 Nov 2023 00:00:00 -0600

Over the years, I have published a few articles outlining my job and approach to my career.

However, through my engagement with early career professionals, I still receive questions regarding my background and responsibilities as Chief Technology Officer (CTO) and Chief Information Security Officer (CISO). These questions often extend to the role of IT and the responsibilities of my team

Therefore, I thought I would share a high-level summary of my background, whilst also including a few examples of key initiatives that I hope provide additional context regarding the role, responsibilities and impact of IT within an enterprise business.

In addition, my LinkedIn profile includes more detail, including links to other public publications, etc.

Background

I was born on the Isle of Wight (UK), where I completed my primary education, including a two-year engagement with Cisco Systems as part of their IT and cybersecurity education program. This is where I was first introduced to business-centric technology concepts.

I graduated from Bournemouth University (UK) in 2007, with a 1st Class BSc (Hons) in Business Information Technology. In 2018, I completed my executive education at Stanford University Graduate School of Business (US).

Since 2018, I have been the global Chief Technology Officer (CTO) at Elanco, expanding my role in 2022 to be the combined global Chief Technology Officer (CTO) and Chief Information Security Officer (CISO), reporting directly to the Chief Information Officer (CIO).

In 2023, I was externally recognised as the “Silver Globee” winner at the Globee Awards in the category “Chief Technology Officer of the Year - IT Services”, as well as named a CIO 100 Awards UK 2023 winner, which aims to “recognise the best and brightest technology leaders in the UK”.

I started my career at Eli Lilly, where I joined as a graduate in 2007, following a successful student placement year at Bournemouth University in 2005/2006. I worked at Eli Lilly for twelve years, where I held numerous engineering and architecture roles across Infrastructure, R&D, Commercial and Global Services.

I closed my tenure at Eli Lilly as the Lead IT Architect for Global IT, which covered Emerging Technologies, Cloud, Hosting, Workplace, Networking, Messaging, Collaborations, Service Management, Data, Financial Services, and Data Centres.

Elanco

Founded in 1954, Elanco is a global leader in animal health, providing those who raise and care for animals with products and services that empower them to address global challenges. Elanco has a strong focus on innovation, both in scientific research and daily operations.

Elanco supports two primary markets, specifically:

Pet, covering Pet Owners and veterinarians, promoting the unique and loving bond pet parents share with their companions, and delivering products, services, and insights to ensure every pet can live a long, healthy life.

Farm, covering Beef, Dairy, Poultry, Swine, and Aquaculture, where raising healthy animals limits the potential for disease spread, supports farmers’ livelihoods and provides nutrient-rich meat, milk, fish, and eggs to nourish and support human health.

In summary, Elanco believes healthy animals are key to solving some of the world’s most pressing issues, like food security and nutrition, mental health and social isolation, and environmental sustainability. It all starts with healthier animals!

Elanco IT enables a secure, reliable, and efficient IT ecosystem for approximately 14,000 users (employees/contractors) worldwide, covering approximately 65 physical locations and 18 24x7x365 GxP-compliant manufacturing facilities.

In addition, Elanco IT delivers value through innovative digital products, services, and insights for our customers across 200 brands and 90 countries.

Role and Responsibilities:

Elanco IT is positioned as a market differentiator, delivering value through the creation of innovative digital products, services and insights that help to transform the well-being of animals, people, and our planet.

The role of the Chief Technology Officer (CTO) aims to bridge the gap between business and technology, supporting business development and being accountable for the overarching technology strategy, including investments, architecture, engineering, and external engagement.

Responsibilities include Enterprise Architecture, which is comprised of Business Process, Solution, Data, and Domain architects, covering all Business Functions (e.g., R&D, Manufacturing, Commercial, Global Services, etc.)

The architecture community maintains a framework of authority and accountability that defines and controls the outcomes of information technology, ensuring that prioritised initiatives have the required sponsorship to succeed.

Alongside the architecture community, I provide leadership across engineering, which covers the design and delivery of IT infrastructure, Industrial Control Systems (Operational Technology), and digital products (internal/external).

In addition, I lead the enterprise governance for the use and development of Artificial Intelligence (AI), reporting directly to the Governance Committee, as part of the Board of Directors.

As Chief Information Security Officer (CISO), I lead Information Security and am accountable for cybersecurity governance, risk and compliance, architecture, and operations, covering detection and response, data protection, vulnerability management, and digital forensic analysis.

In partnership with Privacy, Legal, Audit and Compliance, I am accountable for company-wide cybersecurity awareness and education, including regular engagement with Executive Leadership, the Board of Directors, and investors.

Finally, the combined role includes a focus on innovation and external engagement, helping to identify and realise strategic investments that position Elanco as a leader in the animal health market.

Elanco IPO and Bayer Animal Health Acquisition:

In August 2018, Eli Lilly announced that its subsidiary, Elanco Animal Health, had filed a registration statement with the US Securities and Exchange Commission (SEC) for an initial public offering (IPO) of common stock.

As part of the IPO process, I transitioned from my role as Eli Lilly Global IT Lead Architect to Elanco Chief Technology Officer (CTO), accountable for the design and delivery of the IT Vision and Strategy.

Before the divesture, Elanco was completely reliant on IT services from Eli Lilly, covering Identity, Networking, Hosting, Human Resource Management, Enterprise Resource Planning, Productivity, Collaboration, Data, etc.

The divestiture presented Elanco with the unique opportunity to rebuild IT from the ground up, including a new architecture, technologies, processes, and talent.

This unprecedented situation, where the weight of legacy and technical debt was lifted, enabled a “quantum leap” forward through the creation of a modern IT ecosystem.

Simultaneously, in August 2020, Elanco demonstrated its ambition as a standalone company by closing the acquisition of Bayer Animal Health. The transaction, valued at $6.89 billion, positions Elanco as the second-largest independent animal health company globally, unlocking a diverse and comprehensive product portfolio as well as unparalleled service and support.

Considering the scope, time, and budget pressures associated with these large-scale, complex initiatives, alongside the challenges associated with the macroeconomic environment, including the COVID-19 global pandemic, Elanco IT achieved what many external consultants and research analysts described as “impossible”.

Throughout this process, I led the design and recruitment of a world-class IT team, that successfully built an entirely new IT ecosystem, including all foundational capabilities, achieved in approximately 18 months. Delivery highlights include:

Recruitment and onboarding of approximately 270 IT employees across the US, UK, Poland, and India.
Creation and delivery of the IT organisational structure, methodology, processes, etc.
Migrated approximately 14,000 users (Identity, Productivity, Collaboration) during a global pandemic, including a Microsoft Office 365 to Office 365 tenant migration.
Deployed approximately 14,000 devices across 65 countries, following an autonomous (home) deployment methodology.
Migrated approximately 65 physical sites, including 24x7 manufacturing facilities and R&D labs.
Migrated/modernised approximately 1101 applications/services, supporting key business processes.
Enabled SD-WAN globally, exclusively over Internet circuits, significantly reducing network costs.
Achieved 95% cloud adoption, covering SaaS, FaaS, PaaS and IaaS.
Enabled a new Enterprise Data Foundation, enabling data-driven decision-making and driving data-powered products.
Standardised, and migrated to a new Enterprise Resource Planning (ERP) and Human Resource Management (HRM) environment across approximately 45 countries.
Enabled a new approach to Service Integration and Management (SIAM), delivering support efficiencies.

The modern IT Ecosystem established through this journey primed Elanco as a digitally native business, ready to seize the opportunity as a newly independent company.

Business Innovation:

Powered and supported by Elanco IT, Elanco has begun to deliver new innovative products and services to the market across pet and farm.

Elanco has a robust and exciting innovation pipeline, which aims to launch multiple blockbuster products between now and 2026.

Two recent innovation examples include:

Pet: Canine Parvovirus Monoclonal Antibody

Canine Parvovirus Monoclonal Antibody is the first and only USDA-conditionally approved monoclonal antibody (mAb) treatment that targets canine parvovirus (CPV).

While canine parvovirus (CPV) vaccinations are highly effective and widely available, canine parvovirus (CPV) is still a real and ever-present threat to all dogs, with unvaccinated and young puppies the most at risk.

On average, 900 dogs are diagnosed with canine parvovirus (CPV) each day, with an estimated 92% fatality rate if not treated with timely and aggressive supportive care.

Alongside the ground-breaking Canine Parvovirus Monoclonal Antibody treatment, Elanco launched “The Defenders Fund”, with a bold commitment to save one million puppies from canine parvovirus (CPV) by 2030.

Farm: Livestock Sustainability

The world has until 2030 to reduce climate warming or face irreversible harm.

60% of people are getting the wrong nutrition, according to the World Health Organisation (WHO) and Action Against Hunger.

We’re seeing 50% more demand for protein in the next decade than the last.

Livestock sit at the intersection, animals can help be part of the solution to solving some of the most pressing challenges while giving consumers the protein they want.

Elanco has launched multiple first-of-its-kind products and services that reduce cattle emissions, targeting a reduction of up to 40-50%.

The initial focus is methane. Through enteric fermentation, cows generate methane that is released. Methane lasts approximately ten years in the atmosphere, but it is 25x more potent than carbon dioxide at trapping heat – so smaller reductions create a greater impact on temperature.

If the world can reduce methane emissions by just 20-30%, we can make a real impact and reduce climate warming, making climate-neutral farms a possibility this decade.

Alongside the new products and services, Elanco helped create a digital business startup called Athian, which is creating a marketplace within the food system to inset carbon credits and create economic value for farmers and reputational value for protein products by keeping the carbon value within the food system.

The goal is to help ensure Elanco customers can realise the value of their efforts at the farm level.

IT Innovation:

Alongside the discovery, development, and delivery of life science products to support pets and farms, Elanco IT also incubates, powers, and supports digital/data initiatives, enabling improved productivity for employees and customers.

Two recent digital/data innovation examples include:

Generative AI for Business

In early 2023, following the market hype associated with Generative AI, Matthew Bull positioned Elanco as a leader by launching ElancoGPT, a Generative AI chatbot, designed to accelerate productivity by streamlining common business tasks.

ElancoGPT can be compared to ChatGPT, capable of understanding and generating human-like responses.

Uniquely, ElancoGPT is private to Elanco, with specific business features and controls to ensure security, privacy, legal, and compliance.

For example, Identity Access Management (FIDO2 Compliant), Individual Session Management, Custom Personas, Business Data/Context Ingestion, Logging/Monitoring, etc.

Thanks to the established Modern IT Ecosystem and talented engineers, ElancoGPT was developed internally in two weeks, leveraging a decoupled architecture that includes a custom user interface and middleware API. It can connect to multiple large language models, specifically the GPT Family, Google PaLM2/Gemini, Imagen 2, etc.

In addition, to enable autonomy and scale, a feature, like OpenAI GPTs, enables business users to securely create their own Generative AI chatbots (known as a custom persona), primed for a specific use case, with configurable context.

Finally, business data/context is ingested through an embedding engine (Retrieval-Augmented Generation), including the ability to scale for multiple use cases, whilst remaining compliant with our security, privacy, and legal policies.

Following an early adopter phase, ElancoGPT was made available globally to all viable users (approximately 7,000), targeting a wide range of common business tasks. For example:

Answers (Information, Knowledge Articles, Guides, etc.)
Ideation (Brainstorming, Storyboarding, etc.)
Content Creation (Documents, Policies, Processes, Communications, etc.)
Content Review (Literature, Journals, Papers, Press Releases, etc.)
Summarisation (Executive Summaries, Key Messages/Findings, Reports, etc.)
Research Analysis (Qualitative, Observations, Deductive Reasoning, Survey, etc.)
Software/Data Engineering (100+ Programming/Scripting Languages)
Translations (85+ Languages)

Since its launch, ElancoGPT has been used by 5,470 unique users (78% of the viable audience), with approximately 1200 logins per day. That includes 623 active custom personas.

Based on conservative estimates, this usage has driven over 19,000 hours of timesaving, which is the equivalent of $946,424 of savings through productivity enhancements and cost avoidance.

This adoption has been organic (no formal organisational change management) and the service only costs approximately $300 per month to operate (all users), delivering a compelling return on investment.

Sustainable Computing

The world produces more than 50 million tons of e-waste per year. On average, that is 6kg per person, per year.

This e-waste has a direct impact on the well-being of animals, people, and the planet, contributing towards global issues such as climate change.

Supporting the Elanco Healthy Purpose, Elanco IT is continuously exploring new ways to minimise our environmental impact, prioritising partnerships that improve sustainability.

In June 2023, Matthew Bull established a partnership with Framework, launching a pilot that aimed to unlock sustainable computing for enterprise businesses.

The desired outcome is to fundamentally change the model for end-user computing, delivering greater user flexibility and lower costs, whilst reducing the environmental impact.

This model can replace the legacy approach to lifecycle management, whilst also delivering a “near zero waste” process that encourages reuse and longevity.

In partnership with Framework, the pilot, delivered across the UK and US focused on the processes, capabilities, and architecture across the following areas:

Enrolment (Microsoft Autopilot, Intune Compatibility)
Supportability (Repair, Maintenance)
Lifecycle Management (Upgrades, Recycling)
Global Distribution (Value Added Reseller Support)
Security (Hardware, Software Controls)
User Experience (Performance, Usability, Durability)

Results from the pilot have been favourable, highlighting a compelling return on investment, with a potential ten-year total cost of ownership saving of $18M and Co2 equivalent savings of up to 34,000kg.

In November 2023, the pilot was approved for phase one scale, targeting the UK and Europe in 2024.

People Leadership:

Elanco strives to foster the inclusion of all employees, customers, and stakeholders around the world, and celebrate the diversity of its individuals. Elanco unequivocally condemns hate-based violence, which targets people because of their differences, be it disability, race, religion, sexual orientation, or gender identity.

Currently, Elanco IT has approximately 270 full-time employees and 170 contractors. There are also approximately 890 outsourced roles from managed service providers (e.g., Service Desk, Site Support, Application Management, Network Management, etc.)

Approximately 100 of the employees are focused on engineering, architecture and/or cybersecurity, which either directly or indirectly report to me as Chief Technology Officer (CTO) and Chief Information Security Officer (CISO).

I provide leadership, coaching and mentoring across this community, encouraging curiosity, and helping to support individuals’ development/career progression.

I am also a sponsor of the Student Internship Programme, with specific engagement across the UK, US, and India. The goal is to identify STEM talent early, offering individuals an opportunity to learn/develop as part of an industrial placement (lasting up to one year). This programme has proven highly successful, including direct engagement from Bournemouth University, resulting in approximately 45% of all UK IT employees being recruited via this channel, with excellent talent retention.

Finally, I sponsor and mentor the global Elanco Women in Leadership programme and proactively engage the external community, for example, a partnership with the UK Charity Wonderseekers through the Elanco Global Day of Purpose, whose mission is to create a fairer world where all children can treasure science and are ready and able to improve lives and protect and heal our planet.

OpenAI Chaos

Wed, 22 Nov 2023 00:00:00 -0600

In what will likely become “the” technology story of 2023, the past 132 hours at OpenAI have been chaotic, to say the least.

Reddit user “Saffronfan” has done a nice job documenting the events, which I have summarised below for prosperity.

As a reminder, OpenAI became the centre of the technology world in NOV-2022, with the launch of ChatGPT, resulting in an estimated $80 billion valuation.

OpenAI has a complex governance model, where a non-profit board controls a for-profit business. Therefore, OpenAI employees have equity in the for-profit, but the CEO and the directors do not. Investors in the for-profit (e.g., Microsoft) have restrictions on their return and no control rights, which remain with the nonprofit board.

Timeline of Events

Friday (17-NOV-2023)

CEO Sam Altman Dismissed: Sam Altman was unexpectedly removed as CEO, surprising the technology and business communities. The reason for his dismissal was unclear, causing dismay and anger among investors. (Tweet that started it all).
Co-Founder and President Quits: Following Altman’s departure, CTO Mira Murati was appointed interim CEO, while co-founder Greg Brockman quit. Ilya Sutskever the co-founder and chief scientist is perceived to be at the head of this coup.
Support for Altman: Following Altman’s ousting, there was significant support for him from prominent technology figures, investors, and OpenAI staff, with efforts to reinstate him led by Microsoft. He tweeted he loved his time at OpenAI upon being fired.
Brockman’s Timeline: Brockman went to Twitter to provide his “official” timeline of events.

Saturday (18-NOV-2023)

Board to Reinstate Altman: On Saturday (18-NOV-2023), reports claim the board was in discussion to reinstate Altman as CEO.
Board Pressure: Investors and OpenAI employees threatened the board ordering them to resign and reinstate Altman as CEO or they would lose funding and employees respectfully.
Resistance to Reinstating Altman: Despite investor pressure, OpenAI’s board seemed to resist reversing their decision. They had until 17:00 on Saturday (18-NOV-2023) to re-appoint Altman and Brockman but they were silent until Sunday.
OpenAI Love: Brockman tweeted that he loved his time at OpenAI. He received immense support from OpenAI employees. Additional OpenAI employees were fired with little/no notice.

Sunday (19-NOV-2023)

Internal Conflict at OpenAI: The crisis was attributed to differing personalities and ambitions within OpenAI’s leadership, particularly between Altman and Sutskever, a key figure in AI research.
Altman Trolls: Sam went to Twitter, stating “First and last time I ever wear one of these”, with a photo of him holding an OpenAI guest badge.
New CEO: OpenAI did not reinstate Altman as CEO, instead hiring Emmett Shear former CEO of Twitch.
Microsoft’s Involvement: Microsoft’s CEO Satya Nadella announced that Altman and Brockman would join Microsoft to lead a new advanced AI research team. And that all OpenAI employees would have a spot in the new division.

Monday (20-NOV-2023)

Sutskever Speaks: Sutskever tweeted “I deeply regret my participation in the board’s actions. I never intended to harm OpenAI. I love everything we’ve built together and I will do everything I can to reunite the company.”
Microsoft Stock: On the news of Altman and Brockman joining Microsoft, their stock is up almost 2.5% Monday (20-NOV-2023) morning.
Shear (OpenAI CEO) Speaks: Shear went to Twitter and shared his thoughts on the whole situation.
Employees Exodus: 700 of the 778 employees at OpenAI drafted and signed a letter telling the board to resign or they will quit.
Altman Might Return: According to The Verge, Sutskever flipped to support Altman, but two more board members must change their minds for Altman and Brockman to return.

Tuesday (21-NOV-2023)

738 Employees Threatened to Quit: 738 out of the 778 employees at OpenAI signed the letter stating they plan to quit. However, there is speculation of coercion.
Shear to Resign: Less than 48 hours into his CEO position Shear announces that he will quit if the board does not provide substantial evidence as to why Altman was fired.

Wednesday (22-NOV-2023)

Altman Back: At 01:03 AM Wednesday(22-NOV-2023), OpenAI Tweeted that Altman was returning. They also announced the new board Bret Taylor (Chair), Larry Summers, and Adam D’Angelo.
Altman Tweet: Altman quickly took to Twitter and posted his thoughts on the whole situation.
Nadella Tweet: Nadella may have been the biggest winner from these events, with Microsoft asserting their influence over OpenAI. He shared his closing thoughts on Twitter.
Shear Tweet: Shear exist as CEO and tweeted “72 very intense hours of work”.
Sutskever Tweets: Sutskever tweeted “There exists no sentence in any language that conveys how happy I am”.

Conclusion

Following 132 hours of chaos, OpenAI has arguably gone full circle, with Altman back as CEO. However, the ramifications of these events will live on, with a newly established board and additional details emerging regarding the investment/influence of Microsoft.

The exact reason for the initial dismissal of Altman is still unclear, with rumours of concerns regarding his commercial agenda and/or safety. There is even speculation of an innovation “breakthrough” that members of the board felt needed to be stopped, or at minimum, slowed.

Personally, I am not surprised by this outcome. The complex nature of the OpenAI governance was always destined to create conflict, with the two opposing forces (profit vs. non-profit) becoming increasingly difficult to maintain, especially with large multi-billion dollar investors (e.g., Microsoft). Compounding this issue was the small, immature board governing OpenAI, a group of individuals who are experts in their field, but lack the required expertise/experience to navigate business complexities. The words, “Engineering is easy. People are hard” has never been so true.

With that said, I had not expected the implosion of OpenAI to occur in such a dramatic fashion, especially so publically, with the majority playing out on Twitter.

I expect these events will be studied for years to come and I certainly look forward to watching the inevitable Netflix documentary!

Windows 365 Frontline

Mon, 13 Nov 2023 00:00:00 -0600

In 2022, I published an article regarding Windows 365.

In its simplest form, Windows 365 is a Cloud PC (AKA Virtual Desktop Infrastructure) service. However, it has some unique advantages for enterprise businesses, specifically the unified user experience as part of Office 365 and full integration with Microsoft Intune.

At work, we have successfully deployed Windows 365 at scale, which has proven to be highly effective, outperforming equivalent services from VMware and Citrix.

Unfortunately, the initial licensing model offered by Microsoft was not optimised for shift workers (e.g., service desk agents, support teams, etc.) Therefore, we were forced to retain legacy capabilities for these use cases.

Thankfully, Microsoft recently improved this situation with Windows 365 Frontline, which makes it easy and affordable to extend the power of Windows 365 to shift workers, allowing them to securely access their personalised Windows experience on any device without the hassle of sharing physical PCs.

Key advantages include:

Increase efficiency for frontline workers by providing personalised Windows 365 Cloud PCs that retain their apps, data, and settings whenever they sign in on any device, from any location.
Cloud PCs enable Bring Your Own PC (BYOPC) and remote work scenarios, so your frontline workers are not limited by the constraints of sharing physical, on-premises PCs.
To start a shift, employees sign in to Windows 365 from the web or the Windows 365 client. When they sign out, a personalised Cloud PC automatically becomes available for the next employee.

We have been partnering closely with Microsoft on this rollout and my colleague Mike Henry (Executive Director, TechOps) was recently featured by Microsoft as a customer reference.

In summary, we continue to be impressed with Windows 365, with Frontline providing improved flexibility, productivity and cost-efficiency for shift workers.

ollama

Fri, 10 Nov 2023 00:00:00 -0600

Since the release of ChatGPT, the world has become captivated by Generative AI.

At work, my team has been developing a Generative AI chatbot for business. I have documented the development process through a series of blog posts (links below).

However, when working individually and/or testing, it can be useful to have an option to run a large language model, locally (offline, no Internet connection required).

I have been testing two options, LM Studio and ollama. Although LM Studio offers a robust user interface and direct access to models from Hugging Face, I have settled on ollama (for now).

ollama is a very lightweight application (Mac and Linux, Windows coming soon) for running and managing large language models via the command line (e.g., Terminal). It is fast and very easy to use, supporting a wide range of popular models, discoverable via the ollama website.

When using a Mac, I recommend Apple Silicon (M1/M2/M3), with the following memory configuration.

3b models generally require at least 8GB of RAM
7b models generally require at least 16GB of RAM
13b models generally require at least 32GB of RAM
70b models generally require at least 64GB of RAM

I am fortunate enough to have a MacBook Pro with an M3 Max (16C CPU / 40C GPU) and 128GB of memory. However, I generally use 7b and 13b models, which deliver good performance and an “acceptable” resource impact (leaving enough memory for other tasks).

With a model installed locally, it is stored at the following location (macOS):

~/.ollama/models/blobs/

The models can be quite large. For exmaple, the 13b llama2 model requires 7.4GB of storage.

ollama has a wide range of models available. However, I recommend the following three.

llama2: An open source model released by Meta. Trained on 2 trillion tokens.
codellama: A model for generating and discussing code.
starcoder: A code generation model trained on 80+ programming languages.

To get started with ollama, simply download and install the application. You can then interact with the application and/or models using the following commands and options.

Commands

The following commands can be used to interact with ollama.

To start ollama, run the command:

ollama serve

To run a model, run the command:

ollama run <model>

To pass a prompt as an argument, run the following command:

ollama run <model> "<prompt>"

To list all locally installed models, including their details, run the following command:

ollama list

To remove a local model, run the following command:

ollama rm

Options

The following options can be used when running a model.

To set session variables, run the option:

/set

To show model information, run the option:

/show

To exit/stop the model, run the option:

/bye

Conclusion

If you are interested in Generative AI, I would highly recommend ollama for testing, development and learning.

Be sure to keep a close eye on the ollama blog for updates and tutorials, as well as their Reddit community. I would also recommend reviewing the README documentation from Jeffrey Morgan on the ollama GitHub repository.

Framework Business Process

Thu, 09 Nov 2023 00:00:00 -0600

Earlier this year, we launched a pilot focused on sustainable computing, partnered with Framework. The framing, context and goals of this initiative are outlined in the following two articles:

This article will aim to provide an update on the pilot, including details regarding our findings (e.g., principles, processes, asset management, technology, etc.)

As a reminder, the scope of our pilot covers Framework Laptop 13 and Framework Laptop 16, targeting a range of business personas. In the context of the Framework Laptop 13, we have selected the AMD Ryzen series. In our testing, this product delivers better performance and battery life, alongside additional benefits such as an integrated NPU (not overly useful today, but something to consider as on-device AI becomes more prevalent).

Objective

Today, enterprise business laptops (e.g., Dell, Lenovo, Apple) have an “appliance-like” design and manufacturing process. Therefore, if a laptop is “broken”, the entire laptop must be replaced.

The laptop itself is rarely broken beyond repair, usually, it is a single component that has failed. The most common issues are related to the display, keyboard, trackpad or battery.

The same logic can be applied to upgrades and life cycle management, recognising that current business processes require the entire laptop to be replaced every three to four years.

We believe that sustainable computing can fundamentally change the traditional model for end-user devices (specificlly laptops), delivering greater user flexibility and lower costs, whilst reducing our environmental impact.

Principles

To be successful at scale within an enterprise business, the provisioning and servicing processes of any device must be simple. For example, although technically possible, asking a business user to manually service their device would be considered unacceptable (imagine a business executive opening their Framework laptop to replace a faulty trackpad).

As a result, we anchored on five key principles.

The end-user experience must be frictionless, with all servicing requirements managed transparently as a back-office process. Therefore, no end user will need to service (build, repair or upgrade) a laptop, outside of user-swappable components.
The laptop is a collection of components. Asset management occurs at the component level.
No impact on provisioning, endpoint management or security processes, policies and standards. New laptops will be shipped pre-built, and enrolled via Windows Autopilot.
User-swappable components will be shipped. Servicing will occur at specific hub locations (not via a third-party provider).
The concept of life cycle management is removed, extending the usable life of laptops. Laptop maintenance and upgrades occur ad-hoc at a component level.

Although these principles place a heavier emphasis on IT, evidence suggests any resource/cost impact can be mitigated with the appropriate forethought.

Processes

Testing confirms that our current provisioning and de-provisioning processes can remain unchanged (standardised for all laptops). This is made possible thanks to our use of PC-as-a-Service, which incorporates technologies such as Windows Autopilot and Microsoft Intune.

For example, a new pre-built laptop will be shipped by Framework directly to the user (or via a value added reseller) and automatically enrolled using Windows Autopilot. Therefore, the user simply authenticates over the Internet to synchronise their configuration, applications, data, etc.

However, the servicing and/or upgrade process includes new steps that target a “zero waste” outcome, whilst protecting the user experience (time to resolve). The flow below highlights the proposed process.

This process acknowledges the availability of “user-swappable components” (unique to Framework), which cover any component that can be installed without a screwdriver (specific components listed below). These components will be shipped to the user upon request.

Advanced servicing would happen at a hub location, either by shipping the laptop or locally if the user is within a 30-mile radius. If remote, users will be shipped a new (refurbished) laptop to avoid any downtime, with their existing laptop being repaired or dissasembled for use as spare parts (zero waste).

Asset Management

Unlike current laptops, asset management with a Framework will occur at the component level, with each compeonent being associated with a “parent” that represents the laptop itself. In the event of a servicing requirement, the individual component(s) will be replaced, transferring the component’s unique identifier. The diagram below highlights this process, which is facilitated by Framework via their hardware serialisation capabilities.

Only electrical components (e.g., Memory, SSD) will be uniquely tracked via a serial number, with mechanical components (e.g., top cover, hinges) being tracked for stock quantity.

New laptops will be provisioned pre-built (not DIY), with any unused components being maintained at the nearest hub location as stock.

Technical Testing

The Framework target specification includes two laptops, covering “Business Standard” and “Business Performance”.

Business Standard

Framework Laptop 13
AMD Ryzen 5 7640HS 4.9GHz (6C/12T)
16GB DDR5-5600 (2x8GB) Memory
Western Digital Black SN770 NVMe 2280 500GB (~5,15GB/s Read)
AMD Radeon 760M (8C)
AMD Ryzen AI Engine (10 TOPS)
Keyboard
Bezel (Black)
13.5-inch 3:2 Display (2256x1504 @ 60Hz)
1080p 60fps Camera
Wi-Fi 6E and Bluetooth 5.2
Fingerprint Reader (FIDO2 Compatible)
Microsoft Pluton Security Processor (TPM 2.0 Compatible)
60W Power Adapter (USB-C)
2x USB-C, 1x USB-A, 1x HDMI

Business Performance

Framework Laptop 16
AMD Ryzen 7 7840HS 5.1GHz (8C/16T)
32GB DDR5-5600 (2x16GB) Memory
Western Digital Black SN850X NVMe 2280 1TB (7.2GB/s Read)
AMD Radeon 780M (12C)
AMD Ryzen AI Engine (10 TOPS)
Optional: Expansion Module AMD Radeon RX 7700S 8GB GDDR6
Keyboard
Input Spacers (Black)
Bezel (Black)
16-inch 16:10 Display (2560x1600 @ 165Hz VRR/FreeSync)
1080p 60fps Camera
Wi-Fi 6E and Bluetooth 5.2
Fingerprint Reader (FIDO2 Compatible)
Microsoft Pluton Security Processor (TPM 2.0 Compatible)
180W Power Adapter (USB-C)
3x USB-C, 1x USB-A, 1x HDMI, 1x Audio

The table below outlines the components that have been individually tested (serviced/replaced), verifying software integrity, provisioning (Windows Autopilot), endpoint management (Microsoft Intune) and security.

Non-electrical components (e.g., top cover, hinges) were also tested. However, these have no impact on software integrity, provisioning, endpoint management or security.

The average time to service a component is approximately 20 minutes, requiring no specialist expertise.

Conclusion

Overall, we believe we have proven the concept of sustainable computing within an enterprise business as viable.

Our estimates highlight significant savings when compared to traditional laptops (e.g., Dell, Lenovo), especially when considering the total cost of ownership over ten years. The same logic can be applied to sustainability, where we expect to be able to extend the usable life by a minimum of 33%, resulting in substantial CO2 equivalent savings.

As a result, we have begun discussions with Nirav and the team at Framework to scale in 2024.

However, it is important to acknowledge that Framework is still in its infancy as a business, with its primary focus continuing to be the consumer market. Therefore, Framework are not ready for global scale, with limitations regarding availability and distribution. To mitigate any risk to our business, scale will occur alongside traditional life cycle management events, incrementally shifting to the new model (with a robust backup in the event of issues).

Viable personas/geographies will be targeted and prioritized. The goal is to have ~1500 Framework laptops deployed by the end of 2024, and up to ~4000 laptops by the end of 2025.

Exciting times ahead! Thanks again to the team at Framework, specifically Nirav, Chris, and Marissa.

Generative AI - Enhancements

Thu, 02 Nov 2023 00:00:00 -0500

Update - I have published a new article providing an end of year update.

In May, I published an article that outlined the development and implementation of a Generative AI chatbot for business.

Since that date, my team has been iterating on this service, adding features such as custom personas, multiple models, and context injection from business data.

This journey has been documented in the following articles.

I am pleased to report, that we continue to see excellent viral growth across the business.

For example, as of November 2023, we have seen 3688 unique users access the service, with between 600 to 800 logins per day. This has resulted in a total estimated return on investment of $432,288 through productivity acceleration. As previously highlighted, the target audience for the service is approximately 7,000 business users. Therefore, we have achieved approximately 53% adoption, with no formal organisational change management.

The screenshot below highlights the latest real-time dashboard, which is made available to all users and is used to measure our key performance indicators.

As stated in the previous article, we have also been experimenting with integrated Generative AI services, such as Microsoft 365 Copilot. Although we continue to see tremendous potential, these services remain cost prohibative. For example, Microsoft 365 Copilot costs $30 per user, per month. Our custom-developed Generative AI chatbot for business costs approximately $120 per month, for all users. Therefore, the difference is significant, dramatically impacting the total return on investment.

Therefore, our current plan is to only leverage an integrated Generative AI service (e.g., Microsoft 365 Copilot) for specific user personas, where the value is clear and can be achieved. As a result, our Generative AI chatbot for business will continue to be the default for all users.

With this in mind, we continue to sponsor and invest in our internal engineering team, working to release the next series of enhancements.

Custom Personas 2.0

The concept of custom personas has been available since the launch of our Generative AI chatbot for business.

However, we have enhanced this capability, allowing users to create more advanced custom Generative AI chatbots, with specific user permissions, custom prompts and data ingestion.

The screenshot below highlights the new customer persona creation page, which is fast and simple to use (no coding skills or specialist experience required).

A custom persona allows any user to “build” their own Generative AI chatbot for a specific scenario, supporting an individual or team use case.

At the time of writing, we have 514 custom personas created, available to the business.

In a future release, we plan to make these custom personas easier to discover and access via a searchable storefront (with visibility and access controlled by user permissions).

This update brings our custom persona feature in line with the recently announced GPTs capability from OpenAI. It was also gratifying to see OpenAI release a feature that we have been using for many months, reinforcing our internal prioritisation.

Image Generation

Our Generative AI chatbot for business has always supported multiple models, specifically GPT-3.5, GPT-4, GPT-4-32K, and PaLM-2. However, for the first time, we have enabled multimodal capabilities, starting with image generation.

Leveraging GPT-4 and Google Imagen, users are guided through a series of steps (using natural language) to produce a new image. The screenshot below highlights this capability in action.

GPT-4 is used to help guide the user, ultimately producing a better prompt that is passed to Google Imagen. At this point, three images are generated, based on the user requirements (including resolution, etc.)

Although this feature offers some immediate value for specific personas (marketing, corporate communications, etc.) The exciting part is the potential for the GPT-4 guided steps to be replicated for use against other APIs to trigger specific actions.

For example, imagine a scenario where a Generative AI chatbot for business can collect user information regarding a specific task, creating the required input to trigger an action with a third-party system (e.g., ServiceNow, Workday, SAP, Salesforce.com).

Context Ingestion from Business Data

Finally, our context ingestion capability, using a tested and tuned embedding engine has been industrialised, including the ability to scale for multiple use cases, whilst remaining compliant with our security, privacy and legal policies.

The screenshot below highlights our Generative AI chatbot for business responding to a business-specific question.

It is worth noting the embedded links (enhancing usability) and the ability for the user to verify the information via a citation.

We continue to believe context ingestion from business data presents the greatest opportunity for value, accelerating productivity by streamlining common tasks, whilst unlocking insights from previously obscured data.

Conclusion

Overall, we remain excited by the business opportunities associated with Generative AI. We are also encouraged to see these opportunities rapidly translated into value, which is not always the case with “hyped” technologies (we are still waiting for Blockchain to have an impact outside of Cryptocurrency).

At this stage, we plan to continue to iterate on our Generative AI chatbot for business, whilst keeping a close eye on the rapidly evolving market covering commercial and open source players.

Framework Laptop 13 - AMD

Sat, 21 Oct 2023 00:00:00 -0500

As highlighted in my previous article, we recently launched a pilot focused on sustainable computing within an enterprise business, partnered with Framework.

I am personally passionate about sustainability, which triggered the formal partnership based on my role as Chief Technology Officer (CTO) and Chief Information Security Officer (CISO).

The articles linked below, highlight my journey with Framework.

As it stands, we have approximately 40 laptops (UK and US), covering Intel and AMD configurations.

With the AMD Ryzen series embargo now lifted, I thought I would share a few thoughts regarding our use of the Framework Laptop 13 AMD Ryzen series. This is not intended to be a full review. For that, I would recommend watching the excellent Elevated Systems video.

The specification of my Framework Laptop 13 can be found below:

Framework Laptop 13 DIY Edition
AMD Ryzen 7 7840U (8C/16T)
64GB DDR5-5600 RAM
4TB Western Digital Black SN850X NVMe (7.3GB/s Read)
13.5-inch LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

This is the “high-end” configuration, representing the best available performance.

I am a Linux user (specifically Red Hat Fedora). However, our business standard is Microsoft Windows 11.

Therefore, I will split my thoughts into three parts, general (applies to both Linux and Windows) and then some specific observations for each operating system.

General

I have used every version of the Framework Laptop 13, specifically the Intel 11th Generation, Intel 12th Generation, Intel 13th Generation and AMD Ryzen series.

Although the form factor and build quality have remained consistent, I can state without question that the performance and power consumption (e.g., battery life) have improved with each upgrade. Therefore, comparing the Intel 11th Generation (2022) with the AMD Ryzen series (2023) is “night and day”, with the AMD Ryzen series delivering more than double the performance when running any application that is multi-thread dependent and/or graphics intensive.

However, it is the heat, noise and battery life where the AMD Ryzen series shines, delivering upwards of four hours of improved battery life over the equivalent Intel configuration. In my opinion, this is the first Framework Laptop 13 with “all day” battery life, removing the need to consistently carry a power supply.

The AMD Ryzen series can still get hot and loud, just like the Intel configurations. However, this appears to be far less common when executing daily tasks, such as productivity, collaboration, web browsing, media consumption, etc. Instead, the AMD Ryzen series only becomes noticeably hot and loud when under sustained load, which is far more acceptable.

A few additional observations that I have not seen as commonly discussed in the public reviews.

The Framework Laptop 13 AMD Ryzen series leverages DDR5 memory (unlike the Intel series, which still uses DDR4). The first boot after building the Framework Laptop 13 AMD Ryzen series was painfully slow, to the point where I questioned if there was an issue. I later discovered, that this was likely due to the memory “training” requirement of DDR5, which can seemingly take up to 15 minutes (depending on the amount of memory installed). Therefore, prepare to be patient with the first boot and do not immediately assume there is a hardware issue.

Another unique aspect of the AMD Ryzen series is the expansion card configuration. Unlike the Intel series, only certain ports support specific expansion cards, with the wrong configuration potentially impacting usability and/or power consumption. The diagram below outlines the Framework certified port configuraiotn.

Finally, although I am confident the AMD Ryzen hardware is a positive step forward compared to Intel 13th Generation, it would appear the software is less mature, resulting in some reliability issues.

For example, inconsistent behaviour when connecting monitors and using external peripherals via USB and/or Bluetooth. Nothing is fundamentally broken. However, it can sometimes take several attempts for everything to work as designed.

Although I cannot be sure, I suspect these issues are software (firmware/driver) related, something that I have reported to Framework for investigation.

Linux (Red Hat Fedora)

At the time of writing, Fedora 39 is not generally available. Therefore, my testing was with a near-final beta using the 3.03 BIOS for AMD Ryzen.

Fedora 39 beta was installed without error, including native drivers for all components. The only issue I had related to the fingerprint reader, which required me to manually install a driver package, as outlined in the Framework Fingerprint Reader guide on the help center.

The benchmarks below are not perfect (due to the software differences). However, it does highlight the strengths of the AMD Ryzen 7 7840U vs. the equivalent Intel i7-1370P.

As the results show, the AMD Ryzen easily outperforms Intel in all CPU, GPU and memory tests. Interestingly, the Intel does deliver better I/O performance. However, my testing could not confirm if this is due to the underlying architecture or differences in the SSD configuration (AMD = 4TB WD Black SN850X NVMe / Intel = 1TB WD Black SN850X NVMe).

Windows 11

The testing with Windows 11 has also been positive. It should be noted that the AMD Ryzen 7040 series is built for Windows 11. Therefore, may not install/run correctly with any prior version of Windows (including Windows 10).

Outside of the general observations (listed above), the performance on Windows appears to match the results shown for Linux.

The only Windows-specific issue identified relates to the Trusted Platform Module (TPM) and Microsoft Intune, which is only a problem for businesses using Microsoft endpoint management.

In short, Microsoft Intune does not think the laptop has Secure Boot and Encryption enabled, even though Windows 11 confirms that they are. It is our belief this has to do with the TPM failing attestation, which Microsoft Intune requires to ensure the TPM is valid and certified. It could also be a BIOS maturity issue, specifically related to the use of Microsoft Pluton.

This very specific issue should not impact consumer usage and can be mitigated in a business environment through a custom Microsoft Intune policy. This issue has also been reported to Framework for investigation.

Conclusion

Overall, the AMD Ryzen series is without question the best version of the Framework Laptop 13. It is highly perfomant (Windows and Linux) with excellent battery life. The software does not appear to be as mature. However, I would expect this to improve quickly over the coming weeks/months.

As a result, if you are in the market for a Framework Laptop 13 and do not have a specific reason to select an Intel configuration, I would recommend the AMD Ryzen series.

Ultron NUKI

Tue, 10 Oct 2023 00:00:00 -0500

I recently invested in a smart lock, specifically the Ultion NUKI.

I had previously explored the Yale Conexis but was discouraged by the customer reviews, many of which expressed frustration with the setup process and software reliability.

I was also nervous about switching to a smart lock that completely removed the option for a traditional key. For example, with the Yale Conexis, if the integrated battery runs flat, you must use a 9v battery pack connected via an external port to gain entry. This feels like a cumbersome workaround and does not protect against potential software issues, which could also deny entry.

As a result, I was excited to learn about the Ultion NUKI, which on paper met all of my smart lock requirements.

The Ultion NUKI combines the excellent 3-star PLUS Ultion lock, with technology from NUKI that is interoperable with Amazon Alexa, Apple HomeKit, Google Home, Ring Doorbell, AirBnB and their own NUKI app.

In addition, the Nuki smart functionality is only accessible from the interior, allowing you to reuse your existing external door handle.

This approach has two key advantages, it simplifies the setup process (just three screws) and retains the look of your front door, without advertising to the world that you have a smart lock (which could become a target).

In my experience, the installation process should take no longer than 15 minutes and the app setup process was very intuitive. By default, the NUKI app connects via Bluetooth (Wi-Fi is available when using the NUKI bridge), allowing for one-click opening, alongside a range of scheduling and guest access features.

In the first few weeks, the app/lock has not failed me once, with the only minor inconvenience being the noise it makes when locking/unlocking (sounds like a drill as it rotates the locking mechanism). It is however reassuring to know that if the lock were to fail, I could simply switch back to a traditional key.

In my opinion, the only real consideration is the limitation of using the Ultion NUKI with a Multi Point Lock (MPL), which is a common lock type in the UK.

A Multi Point Lock extends the height of the door, with multiple locking points along the edge. The extra locking points strengthen the security by removing the single point of failure (single deadbolt), delivering a tighter end-to-end seal.

The challenge with a Multi Point Lock is that the locking mechanism must be triggered (usually by lifting the handle) before locking. This presents an issue for most smart locks (not just the Ultion NUKI), which have no way of triggering the locking mechanism.

This limitation does not impact unlocking, but anytime the door needs to be locked (via the smart lock or traditional key), the handle must first be raised. Therefore, it is not possible to lock the door remotely, unless someone is available to lift the handle.

My primary use case for getting a smart lock was to avoid having to take my keys everywhere (as my car is also keyless). Therefore, this locking limitation has not caused any real-world impact.

In conclusion, I would highly recommend the Ultion NUKI. I believe it offers the best of both worlds (traditional and smart), whilst offering a simple setup process and reliable smart features.

FIDO2 and Passkeys

Fri, 06 Oct 2023 00:00:00 -0500

Over the past couple of months, there has been a notable increase in advanced phishing attacks, leveraging QR codes to bypass email protections and a technique known as “Adversary-in-the-Middle”.

Adversary-in-the-Middle is very difficult for the user to recognise and circumvents standard cybersecurity protections, including Multi-Factor Authentication (MFA).

The attack is commonly initiated via phishing, specifically an email with an external link (or an embedded QR code that redirects to an external link).
The link opens a phishing website that acts as a ”reverse proxy”, which forwards a legitimate login website to the user. For example, the Microsoft login webpage.
The phishing website (reverse proxy) is hosted on a domain controlled by the adversary, allowing them to closely mimic the URL, as well as present a valid certificate to the user.
The adversary transparently sits between the legitimate login website and the user, allowing them to access all unencrypted data flowing through the proxy. This includes the username and password, as well as the associated session cookie (enabling persistent access).
At the point in which the user approves the Multi-Factor Authentication (MFA) request, the attacker has successfully compromised the account, with the ability to persist access via the stolen session cookie.

Unfortunately, all traditional forms of Multi-Factor Authentication (MFA), including Phone Calls, SMS, Authenticator Apps, and Number Matching are susceptible.

Some controls can reduce the risk of compromise, for example, enforcing compliant devices, etc. However, these controls also reduce flexibility, potentially impacting productivity.

If you are an enterprise business leveraging Microsoft device management and/or cybersecurity capabilities, I would highly recommend reviewing the excellent article from Jeffery Appel, that outlines available Microsoft protections in 2023.

Even with these protections enabled, the goal of every business (small or large) should be to adopt new standards, such as FIDO2 and passkeys.

FIDO2

FIDO2 (Fast Identity Online 2) is, confusingly, the third iteration of the FIDO standard, building upon FIDO U2F and FIDO UAF. It is best known as “the passwordless standard”, emphasizing security, convenience, privacy, and scalability.

In the context of phishing and specifically Adversary-in-the-Middle, FIDO2 employs new techniques that are (at least today) considered to be breach resistant.

FIDO2 leverages standard public key cryptography keys and challenges to verify the legitimacy of the request, providing stronger authentication. Specifically, FIDO2 includes three steps:

During the registration process, the user client device creates a new key pair. It retains the private key and registers the public key with the target service.
Authentication is completed by the client device proving possession of the private key by signing a challenge.
The client’s private key can be used only after they are unlocked locally on the device by the user. The local unlock is accomplished via a convenient and secure action such as biometrics, etc.

FIDO2 has also been designed to protect user privacy. The protocol does not provide information that can be used to collaborate and/or track users across. In addition, any biometric information, if used, remains on the user’s device.

Passkeys

Based on FIDO standards, passkeys are a replacement for passwords that provide faster, easier, and more secure authentication.

In short, passkeys are a replacement for passwords. A password is something that can be remembered and typed, and a passkey is a secret stored on a user’s device, unlocked with biometrics.

Advantages of using a passkey include:

Intuitive: Creating and using a passkey is as simple as consenting to save and use them. No password is required.
Always Unique: By design, a passkey is unique per service (can not be reused).
Breach Resistant: A passkey is only stored on a user’s devices. Relying Party (RP) servers store public keys. Even servers that assist in the syncing of passkeys can never access or use the private keys.
Phishing Resistant: Passkeys can only ever be used to authenticate with the appropriate service, avoiding the need for human verification.

Conclusion

As previously stated, I highly recommend all businesses establish a roadmap to adopt FIDO2 and passkeys. If you are a Microsoft Entra ID customer, the team recently announced support for device-bound passkeys stored on computers and mobile devices coming in January 2024.

As an individual, Bitwarden are expected to release their passwordless compatibility this month (October 2023). This will allow users to store and use passkeys alongside traditional passwords in Bitwarden Password Manager.

In the interim, if you are Bitwarden user, I recommend enabling and using FIDO2 WebAuthn, which will at least protect your vault from Adversary-in-the-Middle attacks.

Day of Purpose

Thu, 05 Oct 2023 00:00:00 -0500

Today, my UK IT team (around 35 people) completed a “Day of Purpose” with the children’s charity, Wonderseekers, who run Winchester Science Centre.

The Day of Purpose is an annual volunteer day that encourages all employees globally to partner with local charities to make a positive impact.

Wonderseekers are on a mission to use science engagement and interactive science experiences to empower children to improve lives and protect and heal the planet. This mission is tightly aligned with our own, specifically our “Healthy Purpose” and the focus on Science, Technology, Engineering and Maths (STEM).

Winchester Science Centre is also local to the Elanco UK office. Therefore, we hope this partnership can help to inspire the next generation of diverse talent.

As part of the day, we completed a range of activities, including site housekeeping (gardening, jet washing, cleaning), a workshop with their IT team, advising on their architecture and repairing/modernising a space exhibit.

I am a huge advocate for the work completed by Wonderseekers and the Winchester Science Centre Thanks to the entire team for the partnership!

Samsung G95NC

Wed, 04 Oct 2023 00:00:00 -0500

Since August 2020, I have been using a 49-inch super ultrawide monitor, specifically the Samsung C49RG90.

Having switched from dual 27-inch 4K monitors to a super ultrawide, I am confident I will never go back to multiple monitors.

In my opinion, the 32:9 aspect ratio of a super ultrawide delivers the best overall experience when interacting with a desktop operating system (Windows, macOS, Linux). The scale of the working area streamlines most productivity tasks, by avoiding the need to constantly switch windows, whilst allowing for the “focus window” to remain centred.

With this in mind, I recently upgraded to the Samsung G95NC Odyssey Neo G9 240Hz Dual UHD Monitor, which is a colossal 57-inch super ultrawide monitor.

The full specifications of the Samsung G95NC Odyssey Neo G9 can be found below.

Model: Samsung G95NC Odyssey Neo G9
Panel Size: 57-inch
Panel Type: VA (Quantum Mini LED - 2,392 Zones)
Panel Curvature: 1000R
Aspect Ratio: 32:9
Resolution: 7680x2160
Refresh Rate: 240Hz
Variable Refresh Rate: AMD FreeSync Premium Pro
Response Time: 1ms (GTG)
High Dynamic Range: HDR10+ Gaming, VESA DisplayHDR 1000
Brightness: 420cd/㎡ (Typical), 1000cd/m2 (Peak)
Interface: 3x HDMI 2.1 (HDCP 2.2), 1x DisplayPort 2.1 (HDCP 2.2)

I have been researching this monitor for several months as it offers a rare combination of features that I have long desired. Specifically, a 32:9 aspect ratio, high resolution, high refresh rate, variable refresh rate, Mini LED, 1000R curve and “forward-looking” interfaces (HDMI 2.1 and DisplayPort 2.1).

No other monitor on the market today can boast this specification. The best alternative would be the Samsung G95SC Odyssey OLED G9, which on paper has the “benefit” of OLED technology, but is 8-inches smaller, lacks DisplayPort 2.1 and includes the obscure Micro HDMI interface. Although OLED is a superior panel type, I would argue it is less suitable for productivity tasks, where there is a high risk of burn-in.

I had planned to wait until 2024 to purchase the Samsung G95NC Odyssey Neo G9. However, an unfortunate accident accelerated my timeline. While making some changes to my PC, the back panel of the case fell against the display. Thankfully, it didn’t crack, but it did leave a frustrating scrape.

This incident, combined with a £300 (13.6%) discount from “All Your Appliances”, pushed me over the edge!

The Samsung G95NC

There is no denying it, the Samsung G95NC Odyssey Neo G9 is massive! At 1327.5mm wide and weighing 15.4kg (without the stand), it requires some serious desk space.

Thankfully, I own the exceptional Secret Lab Magnus Pro XL sit-to-stand desk, which is 1770mm wide and can hold up to 120kg.

I have mounted the monitor on the Ergotron HX with HD Pivot, which is a heavy-duty monitor arm that can in theory hold a monitor that weighs 19.1kg. I can confirm that the monitor arm does hold the Samsung G95NC Odyssey Neo G9. However, it is right on the limit, requiring the full tension to be applied to the “HD Pivot”.

Thankfully, with the monitor mounted, it feels very sturdy (no wobble) and removes the need to use the included stand (which is very large), delivering a minimal aesthetic and freeing up desk space.

The monitor is connected to two devices, specifically my custom-built PC, which includes an AMD Ryzen 3950X and NVIDIA GeForce RTX 3090FE, as well as my 16-inch Apple MacBook Pro M2 Max. The full specification of each device can be found in the article “My Setup (Q3 2023)”.

Both devices are connected via HDMI 2.1, using the included cable and a “3m Stouchi HDMI 2.1 Certified Cable” purchased from Amazon UK. They output the native resolution (7680x2160) with HDR enabled at 120Hz. At the time of writing, neither device can output the full 240Hz at native resolution, although this may become viable for the NVIDIA GeForce RTX 3090FE via a future driver update.

The MacBook Pro scaling is set to “More Space (7680x2160)”, although HDR and the 120Hz refresh rate are available at all scaling levels.

It is a little disappointing that 240Hz is not usable. However, in 2023, the value of running a refresh rate above 120Hz at this resolution is questionable. Therefore, simply knowing it is supported as a future option is good enough for now.

One important note, any device that is not HDMI 2.1 or DisplayPort 2.1 compatible will likely fail to output at these settings. This includes any Apple M1 device, covering the Pro, Max and Ultra variants.

Regarding the quality of the display, it is marvellous. The jump from 109 PPI with my old 49-inch Samsung C49RG90 to 140 PPI with the Samsung G95SC Odyssey OLED G9 is immediately noticeable, especially in text-heavy applications, such as the Terminal, Visual Studio Code, etc.

The same can be said for the colours, which certainly feel like a step up from the Samsung C49RG90, likely thanks to the Quantum Mini LED panel. With that said, even with 2,392 dimming zones, it is still possible to see some blooming when displaying light content on a dark background (it is no match for OLED). Thankfully, as I use the monitor for productivity, this criticism is minor and not something I consider a “real-world” issue.

Another point of interest is the heat generated by the monitor, which is certainly above what I experienced with the Samsung C49RG90. Interestingly, as the monitor heats up (first 20 minutes of use), it causes an unusual side effect, specifically a small popping sound as the rear plastic expands. As I use the monitor for 5+ hours per day (usually with headphones on), this is a minor concern, but not something I have experienced before.

There is no evidence to suggest the heat will cause any long-term issues, but it is something to consider if you plan to use the monitor in a small room (which will heat up quickly). On a positive note, it will likely help in winter, avoiding the need for additional heating.

Finally, I am not a heavy gamer, but I did launch the new Cyberpunk 2077 v2.x update and can confirm it looks and runs great with “variable refresh rate” enabled.

Pushing 7680x2160 pixels is a very big ask for any system, even with modern deep learning and upscaling techniques enabled. Therefore, I lowered the resolution to 5120x1440 (equal to my old Samsung C49RG90), which is less pixels per frame than native 4K. Combined with techniques such as NVIDIA DLSS, the game looked and ran perfectly well at high/ultra settings with Ray Tracing.

In conclusion, I am very impressed with the Samsung G95SC Odyssey OLED G9. In my opinion, it is the best monitor on the market, assuming you have the need and are fortunate enough to have the funds and space to support a super ultrawide.

Analogue Pocket

Sat, 30 Sep 2023 00:00:00 -0500

Analogue is an interesting company. They design and sell custom FPGA game systems, specifically the Pocket, Super Nt, Mega Sg and the Duo.

FPGA is a type of chip called an integrated circuit, which uses programmable logic blocks that can be altered as needed. This allows for the chip to be programmed to recreate the hardware circuitry of classic game consoles.

The advantage of an FPGA is that it can run the original game from hardware, instead of software emulation, delivering a perfect (or near-perfect) representation. When combined with the correct input, an FPGA can also load from the original game media (cartridge or disc), controlled via the original input hardware (controller, joystick, etc.)

As a result, anyone interested in legally maintaining their retro gaming collection will likely target an FPGA console, knowing that the original hardware is likely obsolete and therefore no longer available for purchase.

Anyone interested in FPGA has likely explored the Analogue range of game systems or built their own using the popular MiSTer open-source project.

I like the mission of Analogue (retro games preservation), but I am not always a supporter of their sales practices. For example, Analogue game systems are often “limited run”, which can drive up costs, promoting scalpers. Unfortunately, the evidence suggests that the supply from Analogue is artificially constrained and not a real-world limitation. Therefore, it would appear Analogue use their low availability as a sales tactic to charge more for their products, which I consider a dirty tactic that is ultimately anti-consumer (hurting their customers).

With that said, I commend their design/manufacturing teams, as they make high-quality, beautiful products.

As a child, I played a lot of games, therefore I enjoy the nostalgia of playing and developing games from this era (e.g., Platformers, Top Down Shooters, RPGs, etc.)

I own the Analogue Pocket (RRP $219), which is a portable (handheld) game system that takes inspiration from the original Nintendo Gameboy. Out of the box, the Analogue Pocket is compatible with the 2,780+ Game Boy, Game Boy Colour & Game Boy Advance games, played from the original cartridges. This can be extended further with cartridge adapters for the Sega Game Gear, Neo Geo Pocket Colour, Atari Lynx, TurboGrafx-16 and SuperGrafx.

The Analogue Pocket includes a 3.5-inch that outputs a 1600x1440 resolution (615 PPI). That is 10x the resolution of an original Game Boy, meaning that any content played on the Analogue Pocket looks incredible.

The Analogue Pocket also includes a range of inputs, including a MIDI USB-A, MIDI IN, Analog Sync Cable, and Pocket Link Cable. These ports are all compatible with all original Gameboy, Game Boy Colour and Game Boy Advanced, helping to ensure full game feature compatibility, such as multiplayer, etc.

In addition, the Analogue Dock (sold separately - RRP $99.99) allows the Analogue Pocket to output to a modern TV, with USB, Bluetooth and 2.4g controller support. I use the 8BitDo Pro 2, which works great for retro and modern games.

Alongside the hardware is a range of software tricks, including original display modes, including display characteristics like backlight LCD effects, pixel grid patterns and LCD sub-pixel patterns. Combined with modern features, such as sleep, wake and save states, the Analogue Pocket makes playing retro games more convenient, without impacting the integrity of the original game.

Regarding games, the list of incredible titles is vast. For example, the Game Boy Advance, which includes critically acclaimed games, including many high-quality ports from the Nintendo SNES. For example.

Super Mario Advance 3: Yoshi’s Island
Mario Kart Super Circuit
Legend of Zelda: A Link to the Past with Four Swords
Final Fantasy VI Advance
Metroid: Zero Mission

If you are interested in retro gaming or simply own a retro gaming library that you are hoping to preserve, I highly recommend an FPGA game system. Assuming you can find a good price, I believe the Analogue Pocket is the perfect companion for any passionate retro gamer!

Integrated Generative AI

Sat, 23 Sep 2023 00:00:00 -0500

Following the launch of ChatGPT in November 2022, Generative AI has captured the imagination of many around the world, including the mainstream media.

This “hype” has energised the technology industry, with almost every technology company releasing press materials that attempt to highlight their relevance, expertise and leadership in this new era of innovation led by Generative AI.

For example, Salesforce.com, ServiceNow, Workday, SAP, Veeva, Monday.com, and many others.

Suspiciously, before November 2022, not one of these companies even mentioned Generative AI as a feature or area of interest/investment.

In the context of enterprise business, there are two notable exceptions, specifically Microsoft and Google.

Microsoft, thanks to its deep partnership with OpenAI, has launched Bing AI and a series of Copilot products, with the flagship being Microsoft 365 Copilot, which integrates with Word, PowerPoint, Excel, Outlook, Teams, etc.

Google has its in-house developed Generative AI models, accessed via Bard and Duet AI, which is the equivalent to Microsoft 365 Copilot, integrated with Google Workspace.

Microsoft 365 Copilot and Google Duet AI are both premium add-ons (not included as part of existing licensing), conincidently, both with a list price of $30 per user, per month.

It is unclear how much these services cost to operate, but at this price point, integrated Generative AI features are a serious investment for any enterprise business. For example, enabling Microsoft 365 Copilot or Google Duet AI across an enterprise business with 15,000 users would cost an additional $450,000 per month ($5.4M per year).

If we assume other technology companies (e.g., Salesforce.com, ServiceNow, Workday, SAP) also plan to monetize their Generative AI features, the combined cost could be astronomical, destroying any return on investment, ultimately making it unaffordable for even the richest companies on earth.

As an alternative, in May, my team launched a custom-developed Generative AI Chatbot for business, designed to accelerate productivity by streamlining common tasks.

I have documented the development process in the following articles.

Our custom-developed Generative AI Chatbot can be compared to ChatGPT, capable of understanding and generating human-like responses to text-based inputs, with specific business features and controls to ensure security, privacy, legal, and compliance.

It should be noted that it is not embedded or tightly integrated into a specific suite of applications or services (like Microsoft 365 Copilot and Google Duet AI).

In my opinion, this is both an advantage and disadvantage, as the decoupled architecture allows for multiple models (e.g., GPT-3.5, GPT-4, PaLM2, Llama 2) to be used, with data being easily ingested from multiple sources (no proprietary integrations and/or specific vendor lock-in).

However, some of the advanced features such as automated actions within applications or services (e.g., Word, PowerPoint, Excel) will be more difficult (maybe impossible) to achieve.

As of September, our Generative AI Chatbot for business has scaled to ~3000 unique users, with the architecture costing a total of $120 per month to operate. If we assume a linear cost increase, which would be the worst-case scenario, 15,000 users would cost just $600 per month ($7,200 per year).

That is just $0.04 per user, per month, resulting in a total yearly saving of $5,392,800 when compared to the list price of an integrated Generative AI service such as Microsoft 365 Copilot or Google Duet AI.

At this early stage, it is impossible to know if an integrated Generative AI service offers enough unique value over a custom-developed Generative AI Chatbot, to cover the additional $5M per year price point. What is clear, having multiple integrated Generative AI services is not feasible (at least today, at the current market pricing).

Therefore, we have chosen to proceed with a strategy that selects one integrated Generative AI service targeting specific users (those where the value is clear), supported by a low-cost, but highly flexible custom-developed Generative AI Chatbot for everyone else.

The document below highlights this positioning, including the differences between an integrated Generative AI service and a custom-developed Generative AI Chatbot.

As noted in the document, our selection for an integrated Generative AI service is Microsoft 365 Copilot, as we are a large Microsoft partner, consuming their core productivity, collaboration, data and information security capabilities.

As previously stated, Microsoft is also a leader in this space with the ability to influence the market dynamics, opening the door to third-party integrations, etc. For example, an ability to access data across the Microsoft Graph or via plugins.

We believe this strategy offers the best of both worlds, securely embracing and enabling Generative AI across the business to unlock new insights and productivity acceleration, whilst managing cost, ensuring the implementation is righ-sized as we scale.

CIO 100 Awards UK

Fri, 22 Sep 2023 00:00:00 -0500

Yesterday, I attended the CIO 100 Awards UK 2023, which aims to “recognise the best and brightest technology leaders in the UK”.

The award ceremony took place at the Leonardo Royal Hotel London St Paul’s, with Gaby Roslin as the host and corporate sponsorship from Cisco, Slack, Red Hat, and others.

With 100 technology leaders in attendance, it was great to connect with peers, sharing success stories and learnings.

I was honoured to be nominated and to receive an award, representing my amazing team and our incredible accomplishments over the past year.

Congratulations to all the winners. Thanks for a fun evening, the nomination and the award!

Purpose and Fulfilment

Sun, 17 Sep 2023 00:00:00 -0500

I have maintained a personal blog (LifeinTECH) for approximately fifteen years.

Although publicly accessible, LifeinTECH is really for me. I find writing strengthens my memory and understanding, whilst serving as a location for personal reflection, allowing me to structure my thoughts.

This is evident when reviewing the 360+ blog posts, with the themes tracking my career from engineer to Chief Technology Office (CTO) and Chief Information Security Officer (CISO).

During this time I have written a few articles that specifically cover my career journey, as well as my rhythms and routines.

Recently, I have been reflecting on motivation. Looking to better understand what motivates me and how to ensure that I have the appropriate balance to support longevity.

When speaking with others, it is not uncommon for people to describe “happiness” as their desired outcome. I understand the sentiment of this statement, but believe it is misguided.

Personally, I seek satisfaction, which in my opinion is deeper than happiness, it is about purpose and fulfilment. For example, eating an ice cream can make me happy. However, achieving a hard-fought goal unlocks satisfaction.

Therefore, fully satisfying my potential, both personal and work, is ultimately how I measure my self-worth.

This leads me to a short list of motivators (the things that give me energy) and demotivators (things that take my energy), with the goal of better understanding and managing my emotional intelligence, hopefully setting myself up for success in daily life.

Motivators:

Opportunities: Complex, high-impact situations, where I have accountability and autonomy to influence.
Action: Situations where meaningful progress can be incrementally measured.
Learning: Continuous self-improvement, covering academic, non-academic (emotional, creative, and metacognitive) and physical.
Coaching: Supporting the growth of others, both personal and career.

Demotivators:

Inefficiency: Bureaucratic thinking that delays action without an objective reason.
Negativity: Incessant complaining without action or when related to an uncontrollable/trivial situation.
Commitment: Failing to meet achievable deadlines and/or expectations.
Illogical: Decisions that defy logic or reasoning, contradicting empirical evidence without supporting context.

In short, I aim to prioritise the areas/traits that motivate me, whilst leading by example through my actions and behaviours.

It is not possible to completely avoid the demotivators. However, if I find myself in a situation where I or others are exhibiting these traits, I attempt to influence/shape the situation to a different outcome. With two small children, I have plenty of opportunities to practice these techniques.

In conclusion, if I were to summarise what motivates me, I would I would state:

“I do not seek happiness; I seek purpose and fulfilment. Accomplished through hard-fought goals by setting high standards, practising temperance, and embracing discomfort.”

This philosophy certainly takes inspiration from Stoic virtues and the state/concept of eudaimonia. I recognise this type of self-reflection can be perceived as pretentious and I caution against “over-thinking”, which can (in my experience) lead to some of the demotivating traits.

However, some metacognitive analysis can be valuable, helping to understand the reasons behind certain thoughts/emotions, hopefully leading to pragmatic steps that enable continuous improvement.

My Setup (Q3 2023)

Fri, 15 Sep 2023 00:00:00 -0500

Update - Hardware specification updated to include the Samsung G95NC, Apple MacBook Pro M3 Max, Asus Zenbook S 16 and Fractal Design Refine.

Final Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

My daily driver is a custom-built desktop PC. The full specification can be found below:

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
NVIDIA GeForce RTX 3090 FE 24GB GDDR6X
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Fractal Torrent Compact

The AMD Ryzen 3950X, Samsung 980 Pro and Nvidia GeForce RTX 3090 are several years old, but still considered premium components, delivering high-performance across a range of workloads (e.g., multi-threading, ray-tracing gaming, machine learning, video editing, etc.)

When remote or travelling, my daily driver is an Apple MacBook Pro 16-inch (OCT-2023). This laptop is also used to run Apple-specific software, such as Xcode and Final Cut Pro, as well as software that requires significant memory (local Large Language Models).

The full specification can be found below:

Apple MacBook Pro 16-inch (OCT-2023)
Apple M3 Max (16-core - 12 Performance / 4 Efficiency)
40-core GPU
16-core Neural Engine
128GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
3x Thunderbolt 4, HDMI 2.1, 3.5mm Headphone, SDXC Card Reader, MagSafe 3
16.2-inch Liquid Retina XDR Display (3456x2234 @ 120Hz, 1600nits)

Finally, I use an Asus Zenbook S 16 running Ubuntu for Linux-specific workloads (e.g., Linux Containers, Ethical Hacking, etc.)

The full specification can be found below:

Asus Zenbook S 16
AMD Ryzen AI 9 HX 370 Processor (12C/24T, 5.1GHz Max)
32GB LPDDR5X RAM
1TB NVMe M.2 SSD
16-inch WQXGA+ OLED Touch Display (2880x1800 @ 120Hz)

At home, the desktop PC and Apple Mac Studio connect to a 57-inch Super Ultra-Wide monitor, specifically the Samsung G95NC Odyssey Neo G9 240Hz Dual UHD Monitor.

The full specification of this massive monitor can be found below.

Model: Samsung G95NC Odyssey Neo G9
Panel Size: 57-inch
Panel Type: VA (Quantum Mini LED - 2,392 Zones)
Panel Curvature: 1000R
Aspect Ratio: 32:9
Resolution: 7680x2160
Refresh Rate: 240Hz
Variable Refresh Rate: AMD FreeSync Premium Pro
Response Time: 1ms (GTG)
High Dynamic Range: HDR10+ Gaming, VESA DisplayHDR 1000
Brightness: 420cd/㎡ (Typical), 1000cd/m2 (Peak)
Interface: 3x HDMI 2.1 (HDCP 2.2), 1x DisplayPort 2.1 (HDCP 2.2)

My workspace is kept fairly minimal, thanks to the exceptional Secret Lab Magnus Pro XL sit-to-stand desk. The monitor is mounted on an Ergotron HX with HD Pivot, which is a heavy duty monitor arm that can (just about) support the 15.4 kg weight of the Samsung G95NC Odyssey Neo G9.

As you can see, for peripherals, I primarily use the Logitech MX Mechanical Mini, Logitech MX Master 3S and Insta360 Link. The keyboard and mouse were selected as they include “easy switch”, which makes it simple to toggle between multiple devices.

Additional peripherals include speakers (AudioEngine A1), headphones (Razer BlackShark V2 and Apple AirPods Max), controller (GuliKit KK3 MAX Controller), steering wheel (Logitech G29), virtual reality headset (Apple Vision Pro and Meta Quest 3) and printer (Epson EcoTank ET-2810).

As highlighted in the photo below, my desktop PC is mounted to the desk using the Secret Lab Premium PC Mount, with the PC case door exposed providing easy access to the components.

Accompanying the desktop PC and monitor is an Original Prusa MINI+, which is an open-source 3D printer.

Finally, I use a Fractal Design Refine chair, which balances comfort and personalised ergonomics.

Home Server

I selected an Apple Mac mini (NOV-2018) for the server, thanks to its excellent power efficiency (150W), high-performance I/O and small form factor. The Mac mini also offers flexible video conferencing capabilities covering Apple FaceTime, Zoom, Microsoft Teams, Google Meet, Cisco WebEx, WhatsApp Video, etc.

Apple Mac mini (NOV-2018)
Intel Core i5-8500B 3.0GHz Base / 4.1GHz Boost (6C/6T)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
256GB PCI-e SSD
Intel UHD Graphics 630
2x 500GB Samsung Portable T5 SSD (USB 3.1 Gen2)

The 1256GB of local SSD storage is split across multiple drives for resilience, with automated backups being completed by Carbon Copy Cloner and Cloud Storage via Microsoft OneDrive.

The only peripheral connected directly to the Mac mini is a Logitech Brio Ultra HD Pro webcam, which delivers phenomenal video quality (4K/30fps - HDR), a wide viewing angle (90-degree dFoV) and a surprisingly good stereo, dual omnidirectional integrated microphone.

To support my minimal (hidden-wire) entertainment setup, all of my AV equipment (Sony STR-DN1050 AV Receiver, Sony PlayStation 5, Logitech Harmony Hub) is located in a cupboard under the stairs, with the required cables fed through the wall to the TV. This can be a little inconvenient when looking to insert/change physical media, but thankfully the cupboard offers plenty of space and ventilation.

Framework Business Pilot

Tue, 12 Sep 2023 00:00:00 -0500

As highlighted in my previous article, we recently launched a pilot focused on sustainable computing, partnered with Framework.

At my company, we prioritise our “Healthy Purpose”, which is the platform through which we engage with our employees, communities and industry leaders across more than 90 countries, to provide global, sustainable solutions to make an impact.

With over 20,000 end-user devices (e.g., laptops, desktops, smartphones, tablets) deployed globally, there is a real opportunity to reduce e-waste, whilst also driving cost savings by re-thinking the processes used to support hardware lifecycle management, upgrades and repairs.

The pilot includes 40 laptops (UK and US), initially prioritising our engineering and scientist communities. The specifications of the laptops being tested can be found below, which is comparable to our current standard provided by Dell.

Framework Laptop 13

Intel i5-1340P 4.6GHz Boost (12C/16T)
16GB DDR4-3200 RAM
Western Digital Black SN770 NVMe M.2 500GB (4GB/s Read) SSD
13.5-inch 3:2 LCD Display (2256x1504 @ 60Hz)
2x USB-C, 1x USB-A, 1x HDMI

Framework Laptop 13

AMD Ryzen 5 7640U 4.9GHz Boost (6C/12T)
16GB DDR5-5600 RAM
Western Digital Black SN770 NVMe M.2 500GB (7.3GB/s Read) SSD
13.5-inch 3:2 LCD Display (2256x1504 @ 60Hz)
2x USB-C, 1x USB-A, 1x HDMI

Framework Laptop 16

AMD Ryzen 9 7940HS 5.2GHz Boost (8C/16T)
32GB DDR5-5600 RAM
Western Digital Black SN850X NVMe M.2 1TB (7.3GB/s Read) SSD
AMD Radeon RX 7700S 8GB GDDR6
16-inch 16:10 LCD Display (2560x1600 @ 165Hz - VRR and FreeSync)
3x USB-C, 1x USB-A, 1x HDMI, 1x Audio

Although we will be collecting user feedback regarding the suitability of each Framework laptop, the primary goal is to validate our hypothesis regarding sustainability and cost efficiencies.

Specifically, we estimate that 8,000 Dell laptops would produce 23,200kg (23.2 Tons) of e-waste. We believe this impact could be reduced to 15,465kg (15.47 Tons) of e-waste when using Framework, which is a decrease of 33% per hardware lifecycle. This would be achieved by extending the life of the laptops by reusing components vs. a full laptop replacement.

The same logic should apply to the return on investment, with 8,000 Dell laptops costing approximately $10,000,000. With Framework, through component servicing and upgrades, we believe the 10-year total cost of ownership could be significantly lower, recognising that this time horizon would ordinarily include multiple hardware lifecycle events.

In addition, we see an average of 500 laptop failures per year, resulting in an additional $650,000 investment. The majority of these failures are serviceable components (e.g., keyboards, displays, batteries). Therefore, Framework could reduce our annual repair cost by up to 86%.

With this context, the pilot covers the following areas, looking to ensure Framework (and sustainable computing concepts) are viable for an enterprise business, specifically the processes, capabilities and architecture.

Enrolment (Microsoft Autopilot, Intune)
Supportability (Repair, Maintenance)
Lifecycle Management (Upgrades, Recycling)
Global Distribution (Value Added Reseller Support)
Security (Hardware, Software Controls)
User Experience (Performance, Usability, Durability)

The processes are a critical area, as today our support and lifecycle management processes are designed to replace the entire laptop, which is very simple (low overhead), but inefficient from a sustainability standpoint. With Framework, we hope to extend the life of the laptop by reusing components. Therefore, the processes will need to be evolved, whilst avoiding user impact and/or incurring additional cost/complexity (which would erode the return on investment).

Alongside the processes, we must also consider the impact on our architecture. Our laptop user base is 95% Microsoft Windows (Microsoft Windows 10/11), leveraging Microsoft Autopilot for enrollment. Therefore, a critical aspect of our testing is to understand the impact of hardware changes, recognising that Microsoft Autopilot is reliant on a hardware hash (uniquely generated from the specific hardware).

The first phase of the pilot will likely run through to December, at which point we will decide the next step. However, we have already started discussions regarding the potential for scale.

Overall, I remain incredibly excited by the work we are doing with Framework and am eager to see how the pilot evolves.

A huge thanks to Nirav Patel, Chris Lombardozzi and the team at Framework for the continued partnership!

Generative AI - Context

Wed, 30 Aug 2023 00:00:00 -0500

Update - I have published two new articles, covering our latest enhacements and an end of year summary.

In May, my team launched a Generative AI Chatbot for business, designed to accelerate productivity by streamlining common tasks.

I have documented the development process in the following articles.

In the last article, I outlined our plan and architecture to ingest business data through a technique known as “embeddings”, which groups similar information using floating point numbers.

The goal was to unlock additional value by enabling the Generative AI Chatbot to respond to prompts with specific business context.

For example, Enterprise businesses have massive datasets (structured and unstructured) that can be a challenge to discover and/or consume, impacting speed to value. We hope that by exposing this data via natural language (Generative AI Chatbot) we can mitigate these pain points, whilst also unlocking additional insights from previously obscured data.

Over the past month, we have been testing this concept across a series of end-user pilots, continuously evaluating and tuning the architecture and data. The desired result is to consistently deliver accurate responses, referenced from business data, avoiding hallucinations.

The screenshot below highlights the business context architecture in action.

In this example, we have ingested our Information Security policy directives and standards (20+ documents and 100+ pages).

The Generative AI Chatbot is prompted with a specific question “What is our password policy?”.

The answer provided is pulled directly from an unstructured document, including the source information to allow for user verification (if required).

The right column shown in the screenshot is for testing purposes only, providing additional detail regarding the specific data chunks and embeddings being referenced. This supports the troubleshooting process, allowing users to better understand why a specific response is being provided.

Overall, testing has been very successful. The architecture tuning process has been surprisingly straightforward, focused on ensuring that the ingested data has been chunked and grouped appropriately, covering issues that can occur from overlapping context windows, etc.

We also “industrialised” the architecture ready for scale, including additional features to enforce/maintain information protection, etc. For example, you do not want restricted/sensitive data to be surfaced via the Generative AI Chatbot by accident.

The bigger challenge has been the data itself, specifically when working with unstructured data (e.g., knowledge articles, policies, procedures, processes)

We have discovered that many of our documents are technically correct, but lack the appropriate context, making them difficult to consume (by a human or a machine). We have also identified a few instances where documents contain incorrect (out of date) information and/or conflict with information found in another document.

When working with thousands of documents, written over many years by different authors, these data issues are to be expected.

When reviewed in isolation by a human, the impact of an individual issue is likely very small. However, when combined, without any ability to apply additional context, the response from a Generative AI Chatbot can be misleading (inaccurate, confusing, conflicting).

As a result, the majority of our time has been spent working with the data owners, supported by some new guidelines and templates to ensure the data is optimised for consumption.

Regardless of our future with Generative AI, this process has been useful in identifying and improving our source data.

As we conclude the pilots, we plan to proceed with a limited production launch, targeting a specific business area. This approach will allow us to continue to learn as we scale, balancing the effort/risk/reward.

In addition to the business context, we have also continued to evolve the core production capabilities. The screenshot below highlights the latest web interface, incorporating a wide range of user experience improvements.

We have also added some specific new features. For example:

Upload Files: Directly upload unstructured documents to be used as part of a specific prompt. This removes the need to copy/paste bodies of text.
Model Switcher: Our architecture is decoupled, allowing for multiple models to be used. At this time we have enabled OpenAI GPT-3.5 (via Azure OpenAI) and Google PaLM2 in production, with OpenAI GPT-4 and Meta Llama 2 in non-production. The user can switch models instantly using the model switcher, whilst maintaining all context and features.
Personas: New pre-primed personas to support specific business tasks. These personas will also be used to surface business context, including individual user access controls, meaning personas can be restricted and secured for specific user groups.
Dashboard: Usage dashboard highlighting the holistic statistics and estimated return on investment.

As highlighted by our dashboard (screenshot below), adoption continues to grow, outpacing our expectations.

Since May, we have seen 2,654 unique users leverage the Generative AI Chatbot, which is approximately 38% of the target audience. On average, this includes 600+ daily interactions.

Considering this growth has been organic (no formal organisational change management intervention), combined with the minimal investment to build/run(approximately $120pm total), we are very pleased with the progress and therefore excited to see what comes next!

Generative AI - Embeddings

Tue, 04 Jul 2023 00:00:00 -0500

Update - I have published three new articles, covering our progress regarding the enablement of context injection (embeddings) at scale, an overview of our latest enhacements and an end of year summary.

Over the past couple of months, my team has been developing a Generative AI Chatbot for business, designed to accelerate productivity by streamlining common tasks.

The following articles document the development process, including the approach, features and architecture.

In mid-June, following an early adopter phase, we released the Generative AI Chatbot to all employees and contractors. We did not formally communicate the release, instead relying upon a one-hour virtual “Tech Talk” that was attended by 480 participants.

In the first few weeks, we have seen impressive adoption, with 1134 unique users, resulting in 9022 logins.

As evidence of the engagement, our “Custom Persona” feature has been used to create 1108 specific Generative AI Chatbots, primed with business context to better support specific individual/team outcomes (e.g., R&D, Manufacturing, Commercial, Communications, Legal, Engineering, etc.)

It should be noted, our architecture is multi-tenant, meaning we have a single instance of the service, serving all users/teams, including all custom personas.

The total viable audience is 14,000 users (number of corporate identities). However, when considering the applicable roles, the target audience reduces to approximately 7,000 users.

Therefore, we have seen organic growth to 16.2% adoption, achieved with no formal organisational change management.

To understand the business value, let’s assume each login saved the user a minimum of 10 minutes. That is a total of 1504 hours (63 days) of productivity returned to the business. At an assumed hourly wage of $48, the productivity return equates to $72,192.

The production service (e.g., web app, middleware, model, data) costs just $105.74 per month to operate, delivering an impressive return on investment.

This early success has energised the team to unlock the next phase of value, achieved via context injection at scale.

Context Injection at Scale (Embeddings)

In my previous article, I outlined our strategy to use traditional data tagging, search and prompt stuffing to add business context.

Unfortunately, this strategy has proven less effective than we had originally hoped, with inconsistent responses that (so far) have failed to meet our quality assurance expectations. For example, unrelated and/or irrelevant information is confidently presented to the user (hallucination).

As a result, we are exploring a more complex architecture, leveraging a technique known as “embeddings”, which groups similar information using floating point numbers. The embeddings can then be used to search for relevant information, injecting the response into the prompt.

The “marchitecture” diagram below highlights the “embeddings” design we have implemented.

As highlighted by the diagram, we have implemented a four-phase architecture to inject context, specifically:

Load: Ingest structured and/or unstructured data from source systems.
Transform: Data quality is important. Therefore, depending on the data type, it must be deduplicated, simplified and/or chunked. For example, splitting large multi-page documents into individual pages. We are also experimenting with converting documents to Markdown or MDX, which can help with formatting, identifying links, code snippets, etc.
Embed: Embeddings help ensure relevant information is injected into the prompt. Embeddings take text (from the transform phase), converting it to a vector (list) of floating point numbers. The distance between two vectors calculates their relatedness, essentially grouping similar (relevant) information.
Store: The embeddings are stored in a vector database (for example, pgvector). This database is searched by taking an input prompt, creating an embedding, and completing a similarity search to find relevant information.

We are currently testing this strategy in non-production against a range of data types, targeting our internal knowledge articles, policies, etc. Although the architecture is higher complexity, our initial results have been positive, delivering more consistent responses and better accuracy.

I will post another update once our testing has concluded.

Framework Build

Fri, 30 Jun 2023 00:00:00 -0500

In my last article, I shared a few details regarding a partnership with Framework, exploring how sustainable computing could be scaled within an enterprise business.

To support this initiative, we launched a production pilot, with the first batch of Framework laptops arriving earlier this week.

In the spirit of demonstrating how quickly and easily a Framework Laptop can be built, repaired or upgraded, I asked my eight-year-old son to complete the process. The specification of the device he built can be found below.

Framework Laptop DIY Edition
Intel i7-1370P 5,2GHz (14C/20T)
64GB Crucial DDR4 PC4-25600C22 3200MHz RAM
1TB Western Digital Black SN850X NVMe (7.3GB/s Read)
Intel Iris Xe Graphics
13.5-inch LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

I first built a Framework Laptop in May 2022, specifically a “DIY Edition” Intel 11th Generation.

Over one year later, I am pleased to see Framework has continued to refine the process, with improvements regarding how the laptop and components are packaged. For example, the “DIY Edition” now comes with the Input Cover Kit and Bezel separately boxed, allowing immediate access to the bottom cover, which includes the mainboard and battery pre-installed.

The photos below highlight the build process, starting with the memory (Crucial DDR4 RAM), using the “Memory Replacement Guide”. It is worth noting that each component includes a QR Code, linking directly to the specific guide (a very nice touch).

Next, the storage (Western Digital Black SN850X NVMe SSD), using the “Storage Replacement Guide”.

Finally, the input cover (which includes the keyboard and trackpad), as well as the bottom fasteners. The steps are outlined in the “Input Cover Replacement Guide”.

The entire process took my son around 20 minutes, with minimal support from myself. I did however take the opportunity to provide a short lesson on computer hardware, but he was more interested in getting hands-on (I don’t blame him).

The fact that an eight-year-old can complete the process with minimal guidance is evidence of the customer-centric approach adopted by Framework!

Framework for Business

Thu, 29 Jun 2023 00:00:00 -0500

The world produces more than 50 million tons of e-waste per year. On average, that is 6kg per person, per year. This e-waste has a direct impact on the well-being of animals, people and the planet, contributing towards global issues such as climate change.

If we hope to improve this situation, everyone must take accountability, proactively contributing towards a better, more sustainable future. This includes enterprise businesses, which due to their scale and centralised governance, are well-positioned to lead from the front!

For example, my company (a relatively small enterprise) owns/manages approximately 20,000 end-user devices, covering laptops, desktops, smartphones, tablets, etc.

Over the past decade, thanks to companies such as Apple, end-user devices have been trending towards “appliance-like” design and manufacturing, reducing or removing any ability to repair and/or upgrade.

Therefore, anytime a device is “broken”, the entire device must be replaced. Hopefully, the broken device is recycled appropriately, but the waste (and cost to the business) is still significant.

Ironically, the device itself is rarely broken beyond repair, usually it is a single component that has failed. In the context of a laptop, the most common issues are related to the display, keyboard, trackpad or battery. Each of these components, with the right design, manufacturing and support processes, could be easily replaced.

The same logic could be applied to upgrades, with most enterprise businesses having a life cycle management process that occurs on a three to five year cadence. Does every user need a new laptop every three to five years? I suspect the majority of devices are still perfectly viable, assuming “worn” components (e.g., keyboard, trackpad or battery) could be replaced, combined with some software housekeeping (factory reset, etc.)

Hopefully, the environmental benefits alone are enough to incentivise change. However, I believe there is also a compelling business case. Assume the average cost for each device is $1,300. That is a total of $26M for 20,000 devices, which could easily grow to $50M+ over a decade. Therefore, even if compared against device leasing options, which can flatten the spending curve, the ability to repair and/or upgrade devices could drive dramatic cost savings, improving the return on investment.

Over the past year, I have been personally following the work being led by Framework, a company producing laptops that respect the consumer’s right to repair and upgrade.

Late last year, we started working with Framework to formally establish the required capabilities and processes to support enterprise businesses, covering topics such as global distribution, scale, enrolment, supportability, security, etc.

Specifically, enterprise businesses have unique requirements regarding how devices are managed. For example, we leverage Windows AutoPilot, which offers many advantages when deploying, managing, and retiring Windows devices at scale. However, Windows AutoPilot is dependent on a device hardware hash, which can add complexity when replacing individual components (e.g., repairs and upgrades)

As a result, we have initiated a production pilot (real users, real work), with the first batch of laptops landing in the office yesterday!

We have initially targeted a specific user persona that we know will require minimal support (e.g., analysts, engineers, etc.) This will allow us to design/test our processes, before wider adoption. As a bit of fun, we also got our early adopters to build their laptops, using the “DIY Edition” and excellent guides provided by Framework.

I am incredibly excited by the work we are doing with Framework! I see a tremendous opportunity to contribute towards a better, more sustainable future.

A huge thanks to Nirav Patel, Chris Lombardozzi and the team at Framework for the continued partnership!

Meta Eye Tracking

Fri, 16 Jun 2023 00:00:00 -0500

Last week, I shared my initial thoughts on the Apple Vision Pro announcement.

I highlighted the use of eye and hand tracking as a revolutionary human interface for Virtual Reality / Augmented Reality, delivering a fast and intuitive way of navigating the operating system and applications.

This led me to think about the Meta Quest Pro, which also includes the required hardware for eye and hand tracking.

Hand tracking is available as part of the Meta operating system and is enabled in some applications. However, up to now, I have only seen the eye tracking capabilities used for foveated rendering (not navigation) to help manage the performance impact of demanding games.

This feels like a huge miss from Meta, but also a huge opportunity!

It is also a little surprising that Meta has not chosen to showcase eye tracking more prominently across their first-party software, recognising it is referenced as a unique feature within their documentation.

“If you choose to turn on eye tracking, your headset will analyse images of your eyes to produce a set of numbers that represent estimates of where you’re looking in VR.”

Thankfully, I was not alone in this thinking.

Earlier today, “ThrillSeeker” released a proof of concept application, replicating the Apple visionOS user interface using the Meta Quest Pro, with eye and hand tracking enabled. In addition, he released a video (embedded below) highlighting the development process, accomplished in just two days using Unity.

ThrillSeeker was not the first to build an eye tracking application for the Meta Quest Pro, with examples (Dilmer Valecillios, Fabio914) dating back six months. However, he is the first to replicate the user experience presented by Apple at WWDC 2023, providing a direct comparison.

The application (Thrills Eyetracked Playground) can be downloaded and installed via SideQuest.

This proof of concept, alongside the many I expect to see in the coming weeks/months, validates my suspicion that the combination of eye and hand tracking could be primed to become the default human interface for Virtual Reality / Augmented Reality.

Similar to what they achieved with multi-touch on the iPhone, I credit Apple for taking an existing concept and leveraging their vertical integration (hardware + software) to deliver an experience that pushes the market forward.

I hope Meta takes inspiration from Apple and their community, enabling first-party support for eye and hand tracking across the operating system and their first-party applications.

If nothing else, considering the Meta Quest Pro is $999 ($2501 cheaper than the Apple Vision Pro) and has immediate availability, it could offer a nice public relations boost to shift some of the media/consumer attention away from Apple.

Finally, it begs the question regarding the upcoming Meta Quest 3, which is a lower-cost ($499) headset that unfortunately does not include the required hardware for eye tracking. In hindsight, this omission may have been a mistake but opens the door to a potential Meta Quest 3+, potentially released in time for the holiday season (and still before the Apple Vision Pro).

I would certainly be open to spending an additional $100 for eye and hand tracking, assuming tight software integration.

Generative AI for Business - Update

Tue, 13 Jun 2023 00:00:00 -0500

Update - I have published four additional articles documenting our strategy to enable context injection (embeddings) at scale, a progress update regarding our findings an overview of our latest enhacements and an end of year summary.

Last month, my team launched a Generative AI Chatbot for business use, designed to accelerate productivity by streamlining common tasks.

It can be compared to ChatGPT, leveraging two large language models (specifically OpenAI GPT-3.5 and Google PaLM2), making it capable of understanding and generating human-like responses to text-based inputs.

The key difference from ChatGPT (and equivalent services) is that our Generative AI Chatbot is private, with specific business features and controls to ensure security, privacy, legal, and compliance.

Initially, we launched the Generative AI Chatbot to 100 early adopters, targetting a wide range of common business tasks. For example:

Answers (Information, Knowledge Articles, Guides, etc.)
Ideation (Brainstorming, Storyboarding, etc.)
Content Creation (Documents, Policies, Processes, Communications, etc.)
Content Review (Literature, Journals, Papers, Press Releases, etc.)
Summarisation (Executive Summaries, Key Messages/Findings, Reports, etc.)
Research Analysis (Qualitative, Observations, Deductive Reasoning, Survey, etc.)
Software/Data Engineering (100+ Programming/Scripting Languages)
Translations (85+ Languages)

I am pleased to report the early adopter phase was a huge success, which saw viral growth above the original 100-user scope. One user commented “this Generative AI Chatbot has given me 50% more productivity in my day”, which directly speaks to the value we had hoped to unlock.

The early adopter phase also provided valuable feedback, resulting in six additional releases during the first month, delivering user experience enhancements and response tuning. We also monitored the usage patterns, further informing our policies and processes regarding the use of Generative AI services.

As a result, following discussions with the relevant internal stakeholders, we agreed to open the service to all employees (approximately 14,000).

This included a new user interface (screenshot below), with a focus on custom personas.

The custom personas incorporate prompt engineering techniques that were tuned during the early adopter phase. Each persona targets a specific outcome, including rules that define the task, format and tone, resulting in a more accurate and appropriate response.

We also enhanced the ability for users to create their custom personas, providing the rules and any context as input, resulting in a unique URL that can be used and/or shared with colleagues. This feature has been used to prime the Generative AI Chatbot with business-specific context, as well as a “quick and dirty” way to test the model against specific data sets (e.g., Knowledge Articles, Processes, Journals, etc.)

For example, the screenshot below highlights the “Software Engineer” persona being used to support common development tasks, which places a focus on producing readable code snippets, incorporating any relevant business context.

In the first business day, 450+ unique users accessed the Generative AI Chatbot, with thousands of prompts being submitted. Thanks to our decoupled architecture (outlined in the previous article) we did not see any scale issues, with the total (all users) cost remaining very low (< $50 per month).

It should be noted, by default, we prohibit the use of public Generative AI services (we are testing GitHub Copilot), which likely influenced the strong adoption.

Our next goal is to unlock business data/context, ingested from multiple data sources. This is a complex problem as data integrity becomes paramount, ensuring the responses accurately reflect the business data/context. As a result, although we have an approach working, we have not yet released this capability to production.

The screenshot below highlights our target state, where the Generative AI Chatbot can interpret business data. For example, in the screenshot, a knowledge article is used as the reference, resulting in a natural language interaction that triggers downstream automation.

Like many enterprise businesses, we have massive datasets (structured and unstructured) that can be a challenge to discover and/or consume, impacting speed to value. We hope that by exposing this data via natural language will improve these pain points, whilst also unlocking additional insights to a broader (non-technical) audience.

In conclusion, many technologies are hyped as “the next big thing”. However, I believe that the current hyper-innovation we are seeing with Generative AI, backed by this massive market investment and momentum, has a real chance of being a global disruptor.

Through my peer connections, I believe we are one of the first enterprise businesses to release a Generative AI Chatbot to all users, including business-specific features and security controls.

As to what the future holds, I am unclear. What is clear, businesses that invest wisely and move with agility will have a tremendous competitive advantage.

Apple Vision Pro

Mon, 05 Jun 2023 00:00:00 -0500

Earlier today, at WWDC 2023, Apple announced their eagerly anticipated Virtual Reality (VR) / Augmented Reality (AR) headset, known as the Apple Vision Pro.

Tim Cook (Apple CEO) described the product as “the first computer you look through, not at”.

I believe this statement accurately represents the ambition, but not the short-term reality. For example, the Apple Vision Pro headset is opaque, with your eyes one inch from the two micro-OLED displays.

The fact that Tim Cook describes the Apple Vision Pro as a computer, specifically a spatial computer, is also interesting.

This subtle shift in terminology is classic Apple marketing, signalling that although the Apple Vision Pro may look and operate like a traditional VR/AR headset, any direct comparison should be avoided. Apple would have you believe this is something different, something new and more compelling.

I was not at WWDC, therefore I have not been hands-on with the Apple Vision Pro. However, the last time Apple entered a new market with hardware was the iPhone, back in 2007 (I consider the iPad and Apple Watch to be an extension of the same theme).

Therefore, as a technologist, I could not resist sharing my initial thoughts on this historic announcement.

Hardware

Although Apple was quick to differentiate the Apple Vision Pro from existing VR/AR headsets, the form factor is remarkably familiar. For example, the photo below is my son using the Meta Quest Pro and Microsoft HoloLens 2.

Like the Apple Vision Pro, the Meta Quest Pro and Microsoft HoloLens 2 are devices you wear on your head (strapped to your face), with your eyes inches from a pair of displays, supported by an array of sensors and cameras.

Before the announcement, I had hoped Apple had found a way to innovate on the form factor, delivering a more streamlined headset that would improve comfort, whilst making it more acceptable/viable to wear across a range of scenarios.

Unfortunately, the “ski goggles” design of the Apple Vision Pro is overly familiar, a continuation of the theme established by the Oculus Rift DK1, released in 2013.

With that said, even from the photos and videos, I am confident the Apple Vision Pro will be the “best” headset on the market. The design, premium materials and subtle queues pulled from other Apple products (e.g., the Digital Crown) all look very refined.

The specification is also a clear generational leap, with two Micro-OLED displays, providing an estimated resolution of 3800x3000 (3400 PPI) per eye and a refresh rate of up to 96Hz. Combined with the M2/R1 architecture, twelve cameras, five sensors and six microphones, the Apple Vision Pro should easily outperform the competition.

There is even an outward-facing lenticular OLED display, designed to indicate to others (not wearing the headset) if the wearer is available to chat or immersed in a virtual experience. This is achieved through a slightly disconcerting virtual representation of the individual’s eyes, which Apple calls “EyeSight”.

Although I applaud the creative thinking, I am not sure this feature is enough to make wearing the Apple Vision Pro socially acceptable. I simply can not imagine interacting with family/friends or co-workers in the real world whilst wearing a bulky headset that obscures half my face.

Finally, Apple was careful not to draw too much attention to the connected battery pack, which is rated for two hours of use. Unlike other VR/AR headsets, the battery back is connected via a cable, which can be kept in a pocket or connected via a belt loop.

The advantage of this approach is that it removes weight from your head. However, I suspect the cable and the need to carry an extra component will quickly become a point of frustration.

The two hours of battery life is also potentially limiting, although it does appear the battery pack includes a USB-C port for additional power.

At this point, with the impressive hardware specification in mind, it is worth mentioning the price. Unsurprisingly, these cutting-edge components and high-quality materials do not come cheap.

The Apple Vision Pro will start at $3,500, which will likely translate to £4,000. To put this into perspective, you could purchase a 13-inch M2 MacBook Air ($1099), iPhone 14 ($799), iPad 256GB ($599), Watch Series 8 45mm ($749), and AirPods Pro ($249) for under the price of the Apple Vision Pro.

It is also significantly more expensive than the competing VR/AR headsets. For example, Meta Quest 2 ($299), Meta Quest 3 ($499), Meta Quest Pro ($999), etc.

The pricing is more in line with enterprise business VR/AR headsets, such as the Microsoft HoloLens 2 (also $3,500). However, this does not appear to be the market being targeted by Apple.

Considering this is an unproven, first-generation product, with a limited third-party ecosystem, the price is ambitious, even for the most “die-hard” Apple fan.

Software

Apple showcased a new operating system known as visionOS (internal codename xrOS). The best equivalent would be iPadOS, with familiar design elements, presented in three dimensions.

The demos of visionOS remind me of the operating system used by the Meta Quest. With infinitely resizable windows, capable of presenting traditional 2D graphics, alongside immersive 3D graphics.

I am sure Apple have some “tricks” not yet revealed, but I was a little surprised not to see a “killer app” that showcased the power of visionOS. Instead, we were treated to a range of demos that were either pre-rendered (representative) or “floating” versions of the apps we use on iPadOS, offering no obvious unique value.

Thankfully, I am confident that Apple will deliver higher quality assurance than the competition. One of my biggest frustrations with the Meta Quest software is how buggy it can be, with frequent pauses, errors and long loading times. This extends into their first-party apps, which should showcase the platform, but commonly act as a point of ridicule (e.g., Meta Horizon Worlds, etc.)

Human Interface

At this point, I would argue that the hardware and software are an evolution, not a revolution. For example, I think you can draw a fairly clear line from the Oculus Rift DK1 (2013) to the Apple Vision Pro (2023).

In fact, many of the concepts enabling the Apple Vision Pro remind me of my first experience with virtual reality, dating back to the early-90s, with the Virtuality 1000CS. This is not necessarily a bad thing, Apple is rarely first to market, but they do usually deliver a quantum leap forward (a revolution).

There was one part of the WWDC keynote that caught my eye (no pun intended).

The majority of VR/AR headsets on the market today include physical controllers for the human interface, whilst also (in certain scenarios) supporting hand tracking. The Apple Vision Pro does not include any physical controllers, instead, it relies exclusively on eye tracking, hand tracking and voice.

On the surface, this feels like a negative. However, based on the initial feedback from individuals at WWDC, it would appear the combination of eye and hand tracking could be a game changer.

As demonstrated in the keynote, you navigate the user interface by simply looking, and touching your index finger and thumb to select. This has the potential to be very intuitive and fast, with some initial feedback from users describing it as “telepathy”.

In addition, the cameras that detect your hands have a very wide field of view, removing the need to hold your arms up for prolonged periods, which can quickly become fatiguing.

I can not wait to try this feature! It could be a real breakthrough for the human interface.

Value Proposition

Apple presented a range of scenarios highlighting how they believe people will use the Apple Vision Pro, including productivity, collaboration, social, media consumption, gaming, etc.

In short, the keynote indicated that the Apple Vision Pro could replace your computer and home media setup, which is an ambitious statement.

This value proposition is not new, simply an extension of the themes previously presented over the past decade by Oculus, Meta, Microsoft, Google, Magic Leap, etc.

Worryingly for Apple, these scenarios, although interesting as a demonstration, have (up to now) failed to capture the imagination of real-world users. For example, VR/AR headsets have not seen mainstream adoption, instead being relegated to niche use cases and/or technology enthusiasts (like me).

At this point, it is unclear if the Apple Vision Pro (including its heavy cost of entry) offers enough unique value to sway consumer opinion.

Conclusion

I have personally owned many VR/AR headsets, including Google Glass, Oculus Rift, Meta Quest 1/2, Meta Quest Pro, Microsoft HoloLens 2, etc. I love the technology and have had some amazing experiences, but these have always been limited to short, specific use cases that emphasise the unique value of VR/AR.

For example, I have previously written about my experience playing Half-Life: Alyx, which remains one of the only “Triple A” games designed specifically for virtual reality.

In my opinion, this game is a masterpiece, putting forward a compelling reason why virtual reality can offer something truly innovative that goes above and beyond traditional gaming.

With that said, I rarely play Half-Life: Alyx for more than 45 minutes at a time. The experience is amazing, making the effort of wearing a cumbersome headset worthwhile, but only for a short period.

There is no question in my mind, the Apple Vision Pro will be a spectacular “product”, with incredible hardware and software.

Unfortunately, I do not believe this guarantees success, underpinned by one simple question…

Does anyone want a large headset strapped to their face?

In my experience, the current hardware limitations make VR/AR viable for short periods, targeting unique use cases. However, I am less convinced general consumers will accept these limitations when positioned as a device that you are expected to wear all day for personal and/or work purposes.

This is especially true when attempting to replace scenarios where VR/AR offer no obvious value over traditional options. For example, does anyone want to go through the effort of strapping a headset to their face for simple tasks such as browsing photos, watching videos, etc? Does anyone want to interact with computer-generated avatars via FaceTime vs. seeing the actual person?

Finally, it was interesting to me that I have yet to see a member of the Apple team wearing the headset. I respect it is very difficult to deliver a compelling VR/AR demonstration on stage. However, part of me believes Apple marketing was eager to avoid the “meme-worthy” ridicule frequently fired at the Meta team for wearing a headset in public and/or interacting in virtual reality.

This feels like a contradiction, considering Apple is advocating for the fact that wearing the headset in public will be socially acceptable.

In conclusion, I can see a world where the concepts behind spatial computing are the future. However, I fear the current hardware limitations associated with the headset are a significant barrier to mainstream adoption.

Potentially better positioning would have been for the Apple Vision Pro to be released for developers only, providing a foundation to build the spatial computing ecosystem and a more compelling value proposition, ahead of general consumer adoption.

I suspect the positioning, timing and price of this first-generation Apple Vision Pro were controversial topics within Apple, potentially more so than any previous Apple product.

Since 2018, Apple has filed approximately 12,300 patents in the US and spent $97B on research and development. Apple stated that they filed over 5000 patents for the Apple Vision Pro. Therefore, although far from a reliable statistic, Apple may have already spent $40B on this product.

That is a lot of money (even for Apple), which will come with a lot of pressure from leadership, shareholders, etc.

I can not help but question if this pressure to deliver a return on investment negatively influenced the product that was announced earlier today.

Either way, I respect Apple for being bold and can not wait to see how general consumers react to the product upon release in Q1 2024.

Globee Awards 2023

Thu, 25 May 2023 00:00:00 -0500

I am pleased to share that I was recently named a “Silver Globee” winner at the 18th Annual 2023 Globee Awards for Information Technology.

I was nominated (without my knowledge) in the category of “Chief Technology Officer of the Year - IT Services”.

The term “Globee” is a combination of the words “global” and “business”. The Globee Awards were formed to honour organisations of all types and sizes from around the world for their achievements in various business and technology-related categories. The awards were created to provide a platform for recognizing and promoting excellence in industries and sectors that are shaping the future of business.

The judging process involved the active participation of over 600 judges from diverse corners of the globe, representing a broad spectrum of industry experts. The Information Technology World Awards welcomes participation from organizations in the Information Technology and Cyber Security sectors, as well as their end-users worldwide.

In addition, my colleague Chris Keeley was a “Gold Globee Winner” in the category of “Chief Information Officer of the Year - Pharmaceutical”.

Thanks for the nomination and award! I simply represent my team, that have delivered exceptional work over the past few years!

Prompt Engineering

Mon, 22 May 2023 00:00:00 -0500

Last week, I shared our progress in developing a business-specific Generative AI Chatbot, designed to accelerate productivity by streamlining common tasks.

The Generative AI Chatbot includes specific business features and controls to ensure security, privacy, legal, and compliance. Specifically, we have enabled a feature that allows the user to declaratively customise the initial prompt through a technique known as prompt engineering, resulting in a custom URL that pre-loads the desired prompt.

Prompt engineering is a technique that involves designing and optimising prompts to improve the performance of a large language model. Prompts are short pieces of text that are used to guide the model’s output, specifying the task to perform, as well as the desired output format, and any other relevant information/constraints.

Our default prompt stipulates that the Generative AI Chatbot is a business assistant and should attempt to respond professionally, using all available data sources. The prompt engineering feature allows for a new custom prompt to be defined (replacing the default), resulting in a new URL that can be used for more specific, tightly defined tasks.

The custom prompt can be used to target a specific persona. For example, Service Desk (Help Desk), Corporate Communications, Investor Relations, Marketing, Paralegal, Software/Data Engineering, etc. This approach helps improve the accuracy of the responses, whilst ensuring they are consistent with the expectations of the specific persona.

As Generative AI gains popularity, prompt engineering has become a more common topic of discussion. Some believe the skills associated with prompt engineering will become a dedicated role. As a general rule, I disagree, instead, I believe the techniques associated with prompt engineering will need to be common across all roles, helping each individual maximise the value of Generative AI capabilities.

Arguably, this is similar to the competency of browsing the web. Individuals who are very good at searching and extracting information from the web are often considered smart when in reality they have simply honed a skill that unlocks efficiency and effectiveness.

Therefore, I thought I would share a few good reference materials that I use to support my prompt engineering requirements.

In addition to these reference materials, outlined below are some high-level principles to consider when prompt engineering.

Clear Instructions: Provide relevant details, context and constraints, limiting the possible outcomes.
Reference Text: Provide reference text as a basis for the response.
Personas: Ask the model to adopt a specific persona/role, which will shape the tone and style of the responses.
Breakdown Tasks: Keep things simple by stepping through requirements instead of providing them all at once.
Iterate: The quality of the prompt will directly impact the quality of the response. If the first response is not as desired, try again with a different structure or context.

It is also possible to ask the model to provide its chain of reasoning, which is useful when troubleshooting why a response may not be meeting your expectations.

These principles, alongside the best practices outlined in the reference materials, can be used in isolation or combined. In theory, they should help reduce artificial hallucination (AKA confabulation or delusion), which is the term used to describe a confident response that is not justified by the training data.

Generative AI for Business

Wed, 10 May 2023 00:00:00 -0500

Update - I have published five additional articles documenting the go-live of our Generative AI Chatbot for Business, our strategy to enable context injection (embeddings) at scale, a progress update regarding our findings an overview of our latest enhacements and an end of year summary.

Since the launch of ChatGPT in November 2022, Artificial Intelligence (AI) has officially hit the mainstream, unlocking new opportunities and risks.

ChatGPT achieved 100 million active users in just two months. This unprecedented growth convinced Microsoft to reprioritise its business strategy, resulting in a rumoured $10B investment in OpenAI and the announcement of portfolio-wide AI capabilities (e.g., 30+ Copilot products in development).

Since February 2023, there has been an onslaught of AI announcements, with every technology company on earth desperately attempting to convince their customer base that they are “leaders in AI”. Specifically, Generative AI, which is a branch of AI that involves the development of models capable of producing novel and realistic content based on patterns and examples learned from training data.

As a CTO, I have been inundated with messages and meeting requests to discuss Generative AI, with imminent announcements expected from Salesforce.com, ServiceNow, Workday, SAP, and many others.

As a recent example, Google mentioned “AI” 143 times during the two-hour keynote presentation at Google I/O, which is approximately 1.15 times per minute. This should not be a surprise, as Google are arguably the company with the most to lose if Generative AI were to disrupt traditional web usage (specifically web advertisement).

In reality, many of these announcements are companies attempting to capitalise on the AI hype cycle (fear of missing out) and therefore should be treated with caution. For example, if a company is positioning their capabilities as “Private Preview” it is reasonable to assume they are in early development vs. “leaders” in this space.

With that said, as highlighted in my previous article, I do see tremendous personal and business potential regarding the use of Generative AI.

As a result, I have a small development team experimenting with these capabilities for business use, prioritising Microsoft (backed by OpenAI), Google and the open-source community.

In my opinion, Microsoft and Google are the most credible technology companies offering business-specific capabilities for Generative AI (Meta is also doing some interesting work). The primary differentiator is the availability of tools, example projects, and architecture materials to support development, including the relevant enterprise business controls (e.g., security, privacy, compliance, etc.)

However, it is arguably the open-source community that is demonstrating the highest rate of innovation with interesting projects such as Stanford Alpaca, Koala, GPT-J, Hugging Face, etc.

The goal of the development team is to build and learn, helping to mature our understanding and shape our thinking regarding the use of Generative AI.

As a result, we have built a business-specific Generative AI Chatbot that is the equivalent of ChatGPT, providing the same ability to answer queries, create and/or review content, generate reports and summaries, accelerate research activities, translations, and much more.

At this time, we have been using two large language models, GPT-3.5 (specifically gpt-3.5-turbo) and PaLM2 (which powers Google Bard), integrated into a business-centric web application where all prompts and responses are private (no interaction with a public interface).

The screenshot below highlights the user interface used to interact with the Generative AI Chatbot. As you can see, it is fairly intuitive and familiar to anyone that has used ChatGPT (or an equivalent Generative AI service).

Regarding features, we have integrated our company security standards, specifically modern authentication (SSO/MFA), alongside individual session management, allowing users to retain context across multiple sessions. These features ensure all prompts, responses, data and use cases are private, adhering to our company’s privacy, legal and compliance policies.

We have also developed a specific capability to declaratively customise the initial prompt (prompt engineering), resulting in a new URL using the same underlying application architecture.

Finally, and maybe most excitingly from a business perspective, we have unlocked business data. This allows the Generative AI Chatbot to respond with business context, covering policies, processes, knowledge articles, journals, papers, etc.

The ability to ingest business data was a little more complex and required the use of some additional cloud components, specifically search (Google Enterprise Search or Azure Cognitive Search) to facilitate the indexing and retrieval of information from internal data repositories.

With the business data indexed, custom middleware extracts the keywords from the user prompt and matches the output (best effort) with the indexed data. The corresponding sentence (or sentences) from the indexed data are provided as part of the response using natural language, including a link (citation) to the relevant data sources.

Behind the scenes, this approach uses a technique known as “prompt stuffing”, where the matched sentences from the indexed data are added (stuffed) into the prompt for the large language model to use as context when responding. This is a fairly brittle approach (something I expect to improve over time), with limitations regarding the token count.

The accuracy of the response can vary depending on the quality of the user prompt and the precision of the matched data. However, we have found the accuracy can be improved when combined with the previously mentioned prompt customisation feature. For example, by stating the Generative AI Chatbot is a support agent and should only respond using a specific internal data repository (e.g., knowledge articles), instead of all known data sources.

The “marchitecture” diagram below highlights the design we have implemented.

The diagram shows a hybrid architecture with the front-end and middleware running on Google Cloud Platform (GCP), connected to Azure OpenAI Service and Google PaLM2 for the large language model. Google Enterprise Search is used to index the the data. It is certainly possible to build this architecture end-to-end on Microsoft Azure or Google Cloud Platform (GCP). However, we purposely decided to target a loosely coupled architecture to provide maximum flexibility as we test and learn. For example, we have multiple large language models available, specifically GPT-3.5 and PaLM2, with the plan to add GPT-4 soon.

At this time, we have opened our Generative AI Chatbot to 100 early adopters, with the plan to expand to 500 in June. This includes a wide range of personas, covering Manufacturing, Research & Development, Commercial, Corporate Communications, Finance, Legal, IT, etc.

The initial feedback has been very positive, with use cases covering:

Answers (Information, Knowledge Articles, Guides, etc.)
Ideation (Brainstorming, Storyboarding, etc.)
Content Creation (Documents, Policies, Processes, Communications, etc.)
Content Review (Literature, Journals, Papers, Press Releases, etc.)
Summarisation (Executive Summaries, Key Messages/Findings, Reports, etc.)
Research Analysis (Qualitative, Observations, Deductive Reasoning, Survey, etc.)
Software/Data Engineering (100+ Programming/Scripting Languages)
Translations (85+ Languages)

Our Generative AI Chatbot is positioned as a productivity accelerator, not a replacement for human intelligence. Therefore, the individual providing the input and receiving the response is accountable for the accuracy (verification).

Do we expect our Generative AI Chatbot to scale further? Maybe… The market is evolving quickly and we will adapt accordingly. For example, we fully expect to see this architecture pattern “bundled” as a service offering from specific vendors. However, it creates questions about cost (assuming everyone will charge a premium for Generative AI capabilities) and the user experience (multiple discrete Generative AI Chatbots).

Ultimately, this project is an experiment (learning experience) for IT and the business, whilst also an opportunity to drive immediate value through enhanced productivity and cost savings.

Finally, it is worth mentioning that our Generative AI Chatbot was built by two developers (Calum Bell and Charlie Hamerston-Budgen), over ten business days (not dedicated to this project), with sponsorship/consultancy from myself and our friends at Microsoft/Google.

Stevie Awards 2023

Fri, 05 May 2023 00:00:00 -0500

Earlier today, my team was named the winner of a Gold Stevie Award at the 21st Annual American Business Awards.

The team won “Technology Department of the Year”, which is a phenomenal achievement, recognising an incredible few years.

The Stevie Awards honour and generate public recognition of the achievements and positive contributions of businesses and working professionals worldwide. The video below provides additional context.

More than 3,700 nominations from organizations of all sizes and in virtually every industry were submitted for consideration. The winners were determined by the average scores of more than 240 professionals worldwide in the three-month judging process.

A few comments from the judges.

“Splits and mergers are never easy and sometimes organizations can get overwhelmed by the complexity and get into a downward productivity spiral. The fact that this department not only came up with a very comprehensive plan but also executed it, is a sign of a very well-functioning team. The team should be very proud of what they have accomplished.”

“The team’s efforts to rebuild the IT infrastructure and integrate the people, processes, and systems from both companies into a single unified platform are impressive.”

Congratulations to the team and all winners!

Wardley Mapping - Update

Mon, 01 May 2023 00:00:00 -0500

As highlighted in my previous articles (Wadley Maps and Mapping), I am a long-time advocate of Wardley Mapping, which is a methodology to build a strategic map for your business.

In my experience, the process of mapping is one of the most powerful, tangible ways to develop a strategy that includes situational awareness and can be easily understood/communicated.

I first met Simon in 2013 as part of a Silicon Valley tour with the Leading Edge Forum (LEF). This led to a specific engagement with my business, where we used the methodology to define our strategy.

Over the years, I have used concepts from Wardley Mapping to support my strategic thinking and again recently to help position emerging technologies/trends regarding Generative AI.

Recently, Simon partnered with “Amazon Web Services (AWS) Public Sector” to produce a series of short videos (each less than ten minutes) that outline the methodology.

If you can spare ten minutes, I would certainly recommend watching the first part (the introduction) to see if it resonates with you.

AI - Rise of the Machines

Sat, 22 Apr 2023 00:00:00 -0500

In November 2022, the OpenAI research laboratory released ChatGPT, a Generative AI Chatbot that can comprehend and generate human-like text, capable of producing novel and realistic content.

This release moved artificial intelligence into the mainstream, capturing the imagination of many around the world.

The “hype” associated with Generative AI has triggered the usual polarising series of public commentary, with some groups positioning artificial intelligence as the world’s saviour, whilst others see it as an existential threat.

To better understand the realities, it is important to remember that artificial intelligence is not new.

The origins of artificial intelligence date back to the early 1900s, with the first discussions regarding the viability of creating artificial humans. Today, artificial intelligence is an embedded part of daily life, used by billions of individuals every day to support a wide range of tasks (selecting music, getting directions, etc.)

However, up to now, artificial intelligence has been a silent companion, usually embedded as a backing service, not immediately obvious to the individual using it. Individuals working with artificial intelligence directly were usually experts within a given domain, with specific expertise and experience.

Generative AI services like ChatGPT changed this equation, allowing anyone to directly interact with artificial intelligence using natural language (no specific expertise or experience required). This paradigm shift dramatically broadens the viable audience, whilst also making the use of artificial intelligence self-evident.

In short, artificial intelligence has been on a 120+ year journey, with the most recent chapter covering the transformer deep learning architecture first proposed in 2017, leading to the inevitable release of ChatGPT in 2022.

Is this an important milestone for artificial intelligence? yes!
Are the machines rising, as seen in movies such as The Terminator? No.

I am not stating that there is no risk associated with artificial intelligence. As with any technology, it can (and will) be used for good and bad.

However, as it stands today, I believe the opportunities associated with artificial intelligence far outweigh the risks. This is especially true for enterprise businesses, who are under continuous pressure to optimise their value chain, whilst innovating to unlock new customer experiences and business models.

Therefore, I have identified four opportunities and four risks that I believe are most relevant (today) for enterprise businesses considering the use of artificial intelligence.

Opportunities

Data Insights:

Artificial intelligence can analyse large datasets quickly, consistently and cost-effectively, identifying patterns, trends, and correlations that would be impossible for a human to recognise. These insights promote data-driven decisions, whilst predicting outcomes that inform business strategies across the value chain.

Productivity Acceleration:

Artificial intelligence enables the automation of repetitive and common tasks, reducing the risk of human errors, whilst empowering every individual to maximize their potential. This leads to increased efficiency, effectiveness, engagement (moral) and cost savings.

New Customer Experiences:

Artificial intelligence can power always-available, personalised customer experiences, maximising go-to-market strategies, whilst improving engagement, response times and overall satisfaction.

New Digital/Data Business Models:

Artificial intelligence can act as the catalyst and facilitate the creation of new digital/data business models, potentially unlocking new monetisation opportunities.

Risks

Competition:

Artificial intelligence has the potential to disrupt traditional business models, powering new customer experiences and unlocking transformative digital/data business models. Businesses that fail to adapt risk becoming obsolete.

Information Protection:

Artificial intelligence is reliant on data, potentially including data that is categorised as sensitive. Data used or produced by artificial intelligence must be appropriately collected, stored, processed, managed and controlled.

Information Accuracy:

Artificial intelligence may misrepresent information with insights/content that appear convincing but may be incomplete, inaccurate, or inappropriate. This can be due to poor-quality training data and/or bias/manipulation. All insights/content must be verified by an accountable party.

Appropriate Use:

The use of artificial intelligence can have ethical, legal, and societal implications, covering consent, fairness, compliance, intellectual property, and the potential for unintended consequences. The use of artificial intelligence must be a conscious decision, focused on the intended outcome and engaged stakeholders.

Conclusion

I am extremely excited about the future of artificial intelligence. In the context of enterprise business, I believe artificial intelligence is primed to trigger the next wave of digital innovation, unlocking opportunities to disrupt and transform industries.

Therefore, although businesses must consider the risks associated with information protection, accuracy and appropriate use, I believe the biggest risk is not embracing the technology. Businesses that fail to do so, risk becoming obsolete.

Cybersecurity Tabletop Exercise

Wed, 12 Apr 2023 00:00:00 -0500

A Cybersecurity Tabletop Exercise simulates a hypothetical cybersecurity incident, used for planning purposes.

It is an opportunity to practice and learn, helping to ensure that processes, roles and responsibilities are clearly defined and understood.

Cybersecurity threats are more prevalent than ever before, impacting all industries. Adversaries are no longer “stereotypical hackers”, they are highly organised teams, operating as a business.

Therefore, if you are accountable for information security at your business, it makes sense to ensure your teams are thoroughly prepared. For example, you do not want the first time you execute your incident response plan to be during a high-pressure, real-world event.

This is the value of the Cybersecurity Tabletop Exercise, raising awareness, learning, identifying gaps, making mistakes and building confidence in a “safe” environment.

A Cybersecurity Tabletop Exercise can cover technical, business and/or executive processes. As a Chief Information Security Officer (CISO), I recommend regular technical and business exercises, combined with occasional executive engagement, which would likely include CxO-level roles and (if appropriate) the Board of Directors.

When constructing a Cybersecurity Tabletop Exercise scenario, I recommend referencing MITRE ATT&CK, which is a globally-accessible knowledge base of adversary tactics and techniques based on real-world observations.

This knowledge base provides a comprehensive matrix for enterprise businesses, perfect for the development of threat models.

Outlined below is an example scenario, covering the confirmed loss of sensitive (non-public) data, including an extortion attempt. It includes four injects, which are milestones in the story that purposely escalate the scenario. At each injection, the team should simulate the defined processes, documenting all decisions and actions, including the rationale.

It can be useful to have a third-party observer, someone with knowledge of cybersecurity incident management, who can assess the effectiveness and provide feedback.

It is important to remember that the scenario itself is designed to represent a worst-case scenario, therefore it should be accepted (not challenged). This is not always easy for technical thought leaders, who have a natural tendency to challenge the plausibility of the scenario itself. Remember, the value exercise is to ensure that the processes, roles and responsibilities are clearly defined and understood.

NOTE: Every business is different, therefore the scenario should be tailored based on your specific business context, emphasising any areas that you may wish to practice.

Inject 1: Third-Party Notification

Your business is notified of probable attack activity by a trusted third-party managed service provider. A user account credential has been found compromised - one belonging to a contractor associated with your business systems.

The third-party managed service provider investigation findings suggest the user compromise occurred 77 days ago.
High-confidence findings identify social engineering as the intrusion vector - resulting in malware installation on a workstation (endpoint) managed by the third-party managed service provider.
Out-of-date virus signature files delayed detection, allowing an adversary to conduct further reconnaissance and exploit activity from the endpoint.

Inject 2: Investigation

Your business identifies the compromised user has accessed your business systems within the window of compromise.

Initial analysis highlights the user accessed your business systems daily.
No unusual account activity identified via your business detect and respond tooling (e.g., no foreign time zones, geographies, impossible travel, etc.)

Inject 3: Findings

Your business operations identify the suspect user account accessed the following production systems/data assets within the window of compromise:

NOTE: Select example systems with appropriate non-public sensitive data (e.g., employee data, customer data, financial data, etc.)

No specific evidence of data exfiltration (copied/transferred). However, with a large compromise window (77 days) data could have been screen-captured and/or transcribed.
Your business operations confirm (within reasonable doubt) that the compromise has been contained.

Inject 4: Data Leak and Extortion Attempt

Your business receives an anonymous phone call from an adversary acknowledging accountability for the compromise, with evidence being published publicly on the dark web. The adversary requests xxx Bitcoin (select a value that has a material impact on your business) be transferred to a crypto wallet within 72 hours, or the full dataset will be published publicly.

The business operations confirm the published evidence is representative of the non-public sensitive data.
The lost dataset cannot be remote/retrospectively recovered and/or obfuscated (masked).

Final Decisions

At this stage, the scenario concludes with a focus on key decisions. This specific example works well, as it includes a long dwell time (77 days) and ambiguity regarding how much data has been exposed as part of the compromise. This uncertainty forces a risk-based decision to be made.

I recommend the following areas are discussed:

How does the business measure the reputation impact and downstream ramifications of the compromise (e.g., customer confidence, sales, stock price, etc.)?
Does the business engage an external cybersecurity/ransomware/extorsion firm?
Does the business pay the ransom? If so, how?
Does the business engage law enforcement?
Does the business directly communicate with employees/customers?
Does the business trigger a formal cybersecurity breach disclosure (e.g., SEC, etc.)?
Does the business trigger a privacy breach disclosure? If so, which Countries?
How does the business monitor the situation post decisions (e.g., customers, shareholders, credit score, etc.)?
What is the broader communication and crisis management strategy?
What further engagement (if any) should occur with the third-party managed service provider (the origin of the compromise)?

To help understand if a Cybersecurity Tabletop Exercise could be useful for your business. Ask yourself how prepared your business would be to answer these questions.

If there is uncertainty regarding the process or decision authority, I would recommend proceeding.

BizClik Media Interview

Wed, 05 Apr 2023 00:00:00 -0500

Earlier this year, in partnership with WinWire Technologies, I completed an interview with BizClik Media, which is a digital community that provides industry-specific expert insights into the fast-paced world of technology, cybersecurity, targeting executive leaders.

The interview focused on scaling public cloud applications and cybersecurity, specifically enabling animal care through cloud technology and modern application architecture.

It has been published in three magazines, specifically:

In addition, the BizClik Media team created a short promotion video highlighting a few snippets from the interview.

Finally, the interview is also available as a Digital Report.

Thanks to WinWire Technologies and BizClik Media for the opportunity to connect and share my perspective.

Arc Browser

Thu, 23 Mar 2023 00:00:00 -0500

Do we need another desktop web browser? If you had asked me at the start of the year, I would have said no… With Chrome, Firefox, Safari, Edge, Opera, Brave, LibreWolf and hundreds of Chromium-based variants, I would have argued there is something already available for everyone and every requirement.

Therefore, I was highly sceptical when I received an invitation to try Arc, a Chromium-based desktop web browser from a start-up known as “The Browser Company of New York” (AKA The Browser Company).

Although I have jumped between browsers over the years, I have always found myself (for better or worse) returning to Google Chrome. However, it has now been several weeks and I have not intentionally opened Google Chrome once on my Mac. Arc has become my primary browser and, as a result, my most used macOS desktop application.

In my opinion, Arc has succeeded where so many others have failed. For example, when Microsoft announced the Chromium-based version of Edge, I felt there was an opportunity to build a streamlined desktop web browser, with an aggressive focus on user experience. Instead, just like Google, they delivered a vehicle to deliver ads and attempt to up-sell Microsoft services, without offering any tangible user experience improvements over Google Chrome.

At its core, Arc is a Chromium-based web browser. That means it is secure, performant (or performant enough) and highly compatible (responds just like Google Chrome). It also has perfect interoperability with all existing Chrome extensions.

The decision to build on a Chromium-base was a wise choice, helping to ensure Arc did not start life at an immediate disadvantage. In short, all of your favourite websites, web apps and extensions (e.g., developer tools, password managers, spelling/grammar checkers) will work exactly as designed.

So why is Ark better than Google Chrome? In my opinion, the primary differentiator is user experience.

Ark moves the traditional tab bar from the top of the browser window to a sidebar. This sounds like a trivial change, but in a world of widescreen displays, it better utilises the screen real estate, whilst also opening the door to a new workflow paradigm.

The sidebar is split into “spaces”, which are personally customisable groups. For example, I have “Personal”, “Work” and “Development” spaces. Spaces can be accessed by swiping horizontally within the sidebar or by clicking the relevant icon (bottom of the sidebar). Spaces can also be isolated (if required), meaning, if you have a personal and work account for the same website or web app, you can log into both simultaneously (no tedious account switching required).

Within each space, traditional tabs are split into three types, “favourites”, “pinned” and “today”.

Favourites appear at the top of the sidebar and persist across spaces. They are great for websites or web apps that you frequently use regardless of your current workflow. For example, I have favourited YouTube, YouTube Music and Reddit. What is even better, Arc includes an excellent media control centre and picture-in-picture mode, meaning I can watch a YouTube video, whilst still navigating other spaces, websites and web apps.

Below favourites are pinned tabs, which can be individual websites, web apps or grouped into folders. Anything that is pinned is persistent for the space and will always be available, with a quick way to return to the default location. Certain pinned websites (e.g., Google Mail, Google Calendar, Microsoft Outlook) also include special features, such as notifications and quick views. For example, within my “Work” space, I pin business websites such as Outlook Email, Outlook Calendar, Monday.com, Workday, ChatGPT, etc.

Finally, “today” tabs are any tabs you opened that day. They can be easily pinned (if required). However, will automatically be moved to “Archived Tabs” at the end of each day. This archive is always available and can be searched or filtered, so nothing is ever lost.

These relatively simple changes have revolutionised my web browsing workflow. Instead of having a long, horizontal row of illegible, co-mingled tabs, covering personal, work and development. I now have structured spaces, which can be switched instantly and always retain the relevant context.

Arc comes pre-configured with robust security and privacy controls (including the ad-blocker uBlock Origin), as well as a few unique features such as Split View, Easels, Notes, and Screen Capture, which further facilitate multitasking and collaboration. To be honest, I have not taken advantage of these capabilities (yet), but as far as I can tell, they are very well-integrated and are not overly obtrusive.

In summary, Arc is the first browser since Firefox was first released in 2004 that has turned my head and retained my attention. It takes all the best parts of Google Chrome and restructures the user experience to better suit modern web workflows. It is not rocket science and I am honestly a little confused about how others (I am looking at you Microsoft) did not see and/or seize this opportunity. However, I am grateful for this outcome, as I would personally prefer to see an operating system-independent browser become successful.

Additional information regarding The Browser Company and Arc can be found linked below. I recommend their YouTube channel, which offers great insight into the team, values, priorities and progress.

My one concern would be how well The Browser Company can scale Arc, whilst retaining what makes it great. The team will undeniably be under pressure to monetise the application, which is a necessary, but dangerous path, frequently leading to shareholder-centric features vs. user-centric features.

It is also worth reiterating that what makes Arc great could be fairly easily copied by one of the other big players (Google, Microsoft, Mozilla), potentially killing Arc before it has a chance to gain meaningful market share.

Finally, Arc is currently only available for Apple macOS, with Microsoft Windows and Linux versions promised. There is also talk of a companion mobile app for Apple iOS, which will be interesting considering the inherent browser limitations of that platform.

If you spend a lot of time using a web browser for disparate workflows (personal, work) I highly recommend you check out Arc. It is currently in beta (invite only).

TechPoint MIRA Awards 2023

Sat, 18 Mar 2023 00:00:00 -0500

Update - Always to bridesmaid, never the bride! Unfortunately, we did not win the main prize. However, we appreciate the external recognition and the opportunity to engage the community. Congratulations to all the winners.

In 2022, my team were nominated and selected as a winner for two TechPoint MIRA awards, specifically “Large Enterprise of the Year” and “Service Partner of the Year”.

The TechPoint Mira Awards are the largest, best-known technology awards for companies operating in the state of Indiana. More than 2,000 nominees have been selected since 1999, with a total of 292 winners.

This year, I am pleased to report the team were nominated and selected as a winner for “Tech Team of the Year”.

This category aims to recognise teams within companies or service providers that have achieved a significant milestone in product development, design, discovery, engineering, company advancement or other through demonstrated problem-solving and/or innovation with measured outcomes.

The winners of 2023 TechPoint MIRA Awards will be announced on Saturday, 22nd April 2023. Wish us luck!

Godot 4

Wed, 01 Mar 2023 00:00:00 -0600

In 2021, I developed a game (Ninja Caves) using the cross-platform game engine GameMaker Studio 2 (now simply known as GameMaker), written in GML, targetting PC with controller support.

The project started as a thirty-day coding challenge, intending to create a simple platformer game (inspired by my childhood favourite Crystal Caves) that my son (six years old) could play.

I documented my progress across three blog posts:

The challenge went well, providing a fun incentive to get “hands-on keyboard” for a personal project, something that I could share with my son.

As a result, I continued the project beyond the original thirty days, eventually releasing the game for Windows, macOS, and iPadOS.

Earlier today, the Godot team released the highly anticipated Godot 4.

Godot is a free and open source 2D and 3D game engine released under the MIT license, which has been growing in popularity over the past few years. Godot offers a lightweight, cross-platform (Windows, MacOS, Linux, BSD), fully integrated game development environment, comparable to the Unreal Engine, Unity and Game Maker Studio 2.

Godot supports a variety of programming languages, including the integrated language GDScript, C++ and C#. GDScript is most interesting to me, as it is a high-level, dynamically typed programming language which is syntactically similar to Python.

Godot 4 is a major release, covering a wide range of new features. The video below from GDQuest provides a great overview.

I have personally been waiting for this release as a reason to dive into Godot. Therefore, I have decided to port Ninja Caves from GameMaker (GML) to Godot 4 (GDScript). The hope is to use this as a mechanism to learn Godot, building on a familiar foundation.

This is very much a side project (hobby). As a result, progress will likely be very slow. However, I do plan to post occasional updates regarding my learnings, with the hope that I can eventually make the Godot version of Ninja Caves available for download.

WinWire - Speed to Value

Tue, 28 Feb 2023 00:00:00 -0600

WinWire Technologies is a digital engineering company that supports enterprises across multiple industries. WinWire has expertise across a range of digital technologies and delivers large enterprise solutions leveraging cloud, artificial intelligence, machine learning, mixed/augmented reality, Internet of Things (IoT), mobility, security, and UI/UX.

In January, the WinWire team interviewed my colleague Sam Miller and I, highlighting some of our recent accomplishments. The synopsis and video can be found below.

“When Elanco began its transformation journey, starting with the announcement to separate from its parent company, complete an IPO and acquire Bayer Animal Health, they had the unique opportunity to rebuild their IT ecosystem from the ground up.

WinWire, a Microsoft Solutions Partner and Data-driven Digital Engineering Company helped build a technology roadmap as part of Elanco’s digital transformation initiative.

Check out what Matt Bull, CTO/ CISO, and Sam Miller, Sr. Director IT – DataOps at Elanco, have to say about how WinWire has been helping the company realize significant time and cost savings.”

Matt Bull and Sam Miller - Elanco Animal Health from WinWire Technologies on Vimeo.

Alongside the interview, WinWire have published two written case studies.

In April, a second interview with Cyber Magazine will be published, providing additional context from a cybersecurity standpoint.

OpenAI ChatGPT

Wed, 08 Feb 2023 00:00:00 -0600

I am extremely excited about the future of artificial intelligence (AI). In the context of business, I believe AI is primed to trigger the next wave of digital innovation, unlocking opportunities to disrupt and transform industries.

In November 2022, the OpenAI research laboratory released ChatGPT, an AI chatbot, built on the GPT (Generative Pre-trained Transformer) family of large language models, moving artificial intelligence into the mainstream, capturing the imagination of many around the world.

ChatGPT can be applied to a wide range of use cases, specifically, answering questions, creating and/or reviewing content, generating reports and summaries, accelerating research activities, translations, and much more.

I believe ChatGPT (and equivalent services) have the potential to make Augmented Automation a commodity within the business, sitting alongside (and supporting) traditional IT Automation and Business Automation. The diagram below highlights one example of how I see AI chatbots (backed by a large language model) being positioned to enable new value.

Earlier today, Microsoft revealed their plan (backed by their significant OpenAI investments) to further integrate OpenAI capabilities across their core services, covering search, productivity, collaboration, software development, etc.

I highly recommend watching the interview below with Satya Nadella (Microsoft CEO). He speaks openly about Microsoft’s strategy, customer opportunities, legal/privacy concerns and market positioning.

It is interesting to see Microsoft take a leadership position regarding these capabilities, especially recognising the risk of a public relations backlash, considering the immaturity of the privacy, legal and compliance regulations covering AI.

With that said, I commend Microsoft for being aggressive and expect their ambition to further accelerate the rate of innovation. For example, this is a clear opportunity for Microsoft to claim market share from Google across search and advertisement. This is big business, striking at the core of the Google product portfolio. Therefore, I fully expect Google to respond, likely accelerating the public accessibility of their in-house LaMDA conversation technology.

This could lead to a direct (and very public) battle between Microsoft and Google, ultimately shaping the future of computing.

As an individual and business leader, this fierce competition is exciting and should result in positive outcomes. Specifically, rapid innovation, competitive pricing, and additional incentives to adopt.

Although Microsoft has led with Bing, as a Microsoft 365 customer, I am most interested in their plan to integrate these capabilities across their core productivity and collaboration suite (e.g., Office 365 and Teams). This would be similar to what they have started with GitHub Copilot, launched back in 2021.

Assuming the capabilities are not locked behind unreasonable licensing, this could be a powerful way of getting millions of users “hands-on” in a very short period. The question for Microsoft is, how do they scale and remain profitable? I assume this would be achieved through careful product/service placement to drive new advertising revenue.

The response from Google, I assume would be similar (e.g., Google Workspace), with a heavier emphasis placed on Google Search integration, capitalising on their dominant 83% market share.

If you are interested in technology, data or AI, I highly recommend you get hands-on! OpenAI ChatGPT and Microsoft Bing AI are available now, with Google’s offering expected in the coming weeks/months.

A word of caution, these technologies are still immature, therefore they may misrepresent information, with responses that appear convincing but are incomplete, inaccurate, or inappropriate. I would also avoid submitting any sensitive information (personal or business) to a public AI service.

In the context of business, I would wait for capabilities such as Microsoft Azure OpenAI Service to mature, which will include the relevant security, privacy, legal and compliance controls.

Exciting times are ahead! I will be watching closely.

External Engagement (JAN 2023)

Fri, 20 Jan 2023 00:00:00 -0600

Update - Both interviews are now live. WinWire Technologies and BizClik Media.

Earlier this month, I completed two interviews, which are both set to be published in Q1 2023.

Each interview builds upon previous external engagement activities (listed below), highlighting specific business outcomes.

The first interview was a customer case study with WinWire Technologies, focused on Hybrid Multi-Cloud, Application Modernisation and Data Insights.

WinWire is a digital engineering company that supports enterprises across multiple industries. WinWire has expertise across a range of digital technologies and delivers large enterprise solutions leveraging cloud, artificial intelligence, machine learning, mixed/augmented reality, Internet of Things (IoT), mobility, security, and UI/UX.

The interview will be published as a short video, where Sam Miller and I provide context regarding our successful partnership with WinWire.

The second interview was with BizClik Media, which is a digital community that provides industry-specific expert insights into the fast-paced world of technology, cybersecurity, targeting executive leaders.

The interview focused on scaling public cloud applications and cybersecurity, specifically enabling animal care through cloud technology and modern application architecture. It will be published in their Technology, Cybersecurity and Healthcare Digital magazines.

I always enjoy engaging with the community and am excited to have these opportunities so early in 2023! I will post an update when the interviews are published.

Meta Quest Casting

Sun, 08 Jan 2023 00:00:00 -0600

I have a Samsung TV (S95B QD-OLD), which includes the ability to wirelessly cast from Apple iOS (Apple AirPlay), Google Android (Samsung SmartThings) and popular apps such as Netflix and YouTube.

Unfortunately, it does not have native support for Google Chromecast, instead requiring a third-party dongle.

This rarely causes an issue. However, the Meta Quest 1/2 and Meta Quest Pro, can only wirelessly cast to the native Meta Quest App or Google Chromecast.

Therefore, officially, a third-party dongle would be required to cast to a Samsung TV, allowing others in the room to view the virtual reality session.

Thankfully, if you have an Apple iPhone or iPad, there is a software workaround.

Download and install the Meta Quest App from the Apple App Store.
Ensure you are signed in on the Virtual Reality Headset and Meta Quest App, with both connected to Wi-Fi.
Select the option to cast from the Virtual Reality Headset to the Meta Quest App.
Select the option to “Screen Mirror” (Apple AirPlay) from the iPad/iPhone, targeting the Samsung TV.

With the iPhone/iPad in landscape orientation, you should now see the virtual reality session appear on the Samsung TV. Quality and latency are not perfect but do improve over time, assuming you have a fast/stable Wi-Fi connection.

The photo below shows my son playing Beat Saber over Christmas.

You can see my iPhone in front of the TV, being used as a casting proxy.

Obsidian

Fri, 06 Jan 2023 00:00:00 -0600

Over the years, I have experimented with many applications for note-taking, including Notational Velocity, Evernote, Microsoft OneNote, Simplenote, Notion, etc.

My approach to note-taking is very simple. I use markdown for everything, which is a lightweight markup language for creating formatted text.

Markdown does not require any special software, making it very accessible, simple to learn, highly portable and human-readable without any translation.

As a result, any plain text editor (e.g., Microsoft Visual Studio Code) can be used to create, modify or view documents written in markdown.

With this in mind, you could question why even use a dedicated note-taking application. In short, specific features that facilitate searching and sorting of notes, which dramatically improve productivity when dealing with hundreds/thousands of disparate files.

Over the past couple of years, I have been using Simplenote. However, I recently wanted to explore an option with a more versatile security model, specifically local (client-side) file encryption.

Outlined below are my primary requirements:

Platform Independent (Windows, macOS, Linux, iOS, Android)
Markdown Support
Cloud Synchronisation Compatible
Local File Encryption Compatible
Tagging Support
Simplified Import/Export
Extensible
Free, Open Source and/or Code Available

Last year, I stumbled upon Obsidian, which was first released in 2020, reaching v1.0.0 in October 2022.

Obsidian

Obsidian is a powerful and extensible knowledge base that works on top of a local folder of plain text files.

At its core, Obsidian is a markdown editor and viewer.

Obsidian stores all notes in a vault. A vault is essentially a collection of markdown files (notes) with some associated metadata files to maintain the user-specific Obsidian configuration.

All files (including the metadata files) are completely portable and can be viewed, modified, copied, and deleted outside of Obsidian. Therefore, there is no need to rely upon proprietary import/export capabilities, as all files can be accessed directly from the filesystem.

Another advantage of this approach is that the vault can be stored anywhere, including your favourite cloud synchronisation service (e.g., Microsoft OneDrive, Google Drive, Dropbox, Nextcloud Files, etc.)

This approach means that Obsidian notes can inherit the security and record retention model associated with these services (versioning, encryption, multi-factor authentication) and/or can be locally secured (local encryption) before cloud synchronisation.

Obsidian also includes features to facilitate searching and sorting. Specifically, a knowledge graph, providing a unique view into the notes, highlighting connections as well as allowing for custom groups, filters, and display options.

The knowledge graph is enhanced through tags, backlinks and outgoing links, which apply additional context to each note, further broadening the view and highlighting connections.

These core features can be extended through a massive array of plugins, supported by a thriving community.

Overall, I have been very impressed with Obsidian and was able to transfer my existing markdown notes with almost no effort. I love the knowledge graph and am excited to explore the plugin options (whilst hopefully maintaining my desired simplicity).

Obsidian is available for Windows, macOS and Linux for free. However, unfortunately, the software is not open source. Thankfully, you can examine the source code (some parts are minified/obfuscated/packed) from the developer console, which is a positive step compared to other closed/proprietary applications.

Password Strength

Sun, 01 Jan 2023 00:00:00 -0600

With the recent LastPass cybersecurity breach, it might be time to reassess your password hygiene.

The table below from Hive Systems (linked from Reddit) estimates the time it would take to brute-force a password in 2022. It should be noted, the time will reduce as computational power increases and/or new techniques are developed.

In short, the minimum acceptable “standard” for a password in 2022 is 16 characters, including numbers, upper and lowercase letters, and symbols. In theory, assuming the password has not been stolen, it would take up to 92 billion years to brute-force.

The table assumes the lowest common denominator for password hashing, specifically MD5. Security-conscious services likely use a stronger hashing algorithm, such as PBKDF2. However, knowing that MD5 is still prevalent, it is good to understand the worst-case scenario. More details regarding the analysis can be found at Hive Systems.

Whilst password strength is important, usability is also critical. Therefore, even with the LastPass cybersecurity breach, I still recommend the use of a password manager. Specificlly, 1Password or Bitwarden.

A password manager enables you to create a unique username/password for every online account that meets or exceeds the previously outlined password guidance (16+ characters, etc.)

As a result, only the username and password of the secure vault used by the password manager are known to the user (master credentials), which are used to unlock (decrypt) the secure vault.

Secure vaults can be hosted privately or via a third party (cloud-hosted). In theory, a third-party hosted secure vault is more vulnerable, as it relies upon the host (e.g. 1Password, Bitwarden) to maintain appropriate security policies, standards, controls, processes, etc.

With that said, a privately hosted secure vault requires more effort to set up and maintain, ultimately making it less convenient/viable for most people.

Therefore, I recommend hosting with a trusted third party (1Password or Bitwarden), whilst ensuring your master credentials (used to encrypt your secure vault) are unique and thoroughly protected.

For example, I recommend the following criteria for your master credentials:

Unique username connected to a personally owned/managed email domain (not @gmail.com)
Unique password
16 characters minimum password length
Password to include numbers, upper and lowercase letters, and symbols
Multi-Factor Authentication (MFA) enabled, ideally a physical key or an authenticator app
Secure vault only accessed via trusted (personally owned/managed) devices

The use of a physical key (e.g. Yubikey) for Multi-Factor Authentication (MFA) is certainly preferred, but not always feasible. Therefore, I would recommend an authenticator app (e.g. Authy), which removes the risk of SIM-jacking attacks and/or common email exploits.

Finally, it is worth checking the type of encryption used by the password manager. If PBKDF2 is used, confirm the default iteration count, which helps protect the master credentials from being brute forced. Bitwarden recommend a minimum of 600,000 iterations (I personally use 1,000,000 iterations).

Will these recommendations ensure your secure vault remains secure? No, nothing can ever be guaranteed. However, for most people, I feel the approach outlined in this article achieves an appropriate balance between security and usability.

Microsoft Account Alias

Fri, 23 Dec 2022 00:00:00 -0600

I registered for my primary email address in 2006 (16 years ago). Since that date, it has been used as the username for countless online accounts.

With hindsight, I would have used a custom domain (something that I own/manage) or created a unique username for each online account using a password manager, such as Bitwarden, etc.

However, in 2006 (I was at University), cybersecurity and online privacy were not high on my priority list.

Therefore, my email address is widely known, having been shared publically thousands of times, either via legitimate channels (with consent) or through one of the many cybersecurity breaches that have occurred during this time horizon.

For example, the website Have I Been Pwned can check to see if your email address has been leaked as part of a known data breach.

Thankfully, modern identity capabilities, such as Multi-Factor Authentication (MFA) and Conditional Access help to mitigate any immediate security concern.

With that said, these identity capabilities will not stop a cyber adversary attempting to gain access to your online accounts via a Brute-force Attack, MFA Fatigue and/or Social Engineering.

A recent example is my Microsoft Account. Looking at my “Sign-in Activity”, I can see my account is under continuous attack.

Thousands of sign-in attempts from unknown source IP addresses, usually via the web, but also using synchronisation options offered by Microsoft (Exchange ActiveSync, etc.)

Although the location field highlights attempts from around the world, the actual origin of the attack will likely never be known, as a cyber adversary will use a range of techniques to obfuscate their true location.

To check your Microsoft Account sign-in activity, navigate to:

My Microsoft Account > Security > Sign-in Activity

As previously stated, good security hygiene and the use of modern identity capabilities will help to ensure the account remains secure. However, this level of continuous attack is a little unsettling.

Therefore, if you find yourself in this position, I recommend considering a change of username (AKA Microsoft Account Alias). Unlike Apple and others, Microsoft makes this process fairly painless and it should immediately remove the risk associated with the original username.

My Microsoft Account > Your Info > Edit Account Information > Account Alias

Here you can add a new account alias (email address). Where possible, I recommend using a secure and privacy-centric email service (like Proton Mail), configured with Multi-Factor Authentication (MFA) and a custom domain.

Once configured (and verified), you can set the new account alias as the primary and (if desired), remove the old account alias. The addition of an account alias may take 24 hours to propagate across all Microsoft services.

Finally, select “Change Sign-in Preferences” and ensure only your new primary account alias is enabled as a sign-in option.

Your account should now be secure, successfully removing the risk associated with the original username. As an additional step, I would also recommend signing out of all existing sessions. To do this, navigate to:

My Microsoft Account > Security > Advanced Security Options

At this point, you should be able to select “Sign Me Out”. This process may take 24 hours to complete across all active sessions.

Unreal Engine

Sun, 18 Dec 2022 00:00:00 -0600

Since January, I’ve been following the work of Reuben Ward (reubs) via his YouTube channel, specifically his series focused on the remake of the game “The Simpsons: Hit and Run”.

The game, “The Simpsons: Hit and Run” was developed by Radical Entertainment (now owned by Activision Blizzard), released in September 2003 for the Sony PlayStation 2, Microsoft Xbox and Nintendo GameCube. At launch, the game received mostly positive reviews but subsequently established itself as a fan favourite.

The remake project began as a “hackathon-style” challenge, where reubs attempted to remake the game in one week using Unreal Engine 5.

The video documenting the process (embedded below) went viral (7.8 million views to date), which prompted the questions… What if you had more than one week? How much of the game could be remade by a single developer?

As of December, reubs has released nine videos, demonstrating unbelievable progress. At this stage, the game is largely playable and includes many “quality of life” improvements, covering performance, graphics, gameplay, artificial intelligence, etc.

If you are interested in game development (or retro gaming), I highly recommend watching this series of videos (playlist linked below). Not only does reubs do an excellent job of documenting his process, which is great for learning new tips/tricks. He also delivers the content with light-hearted commentary, making it a joy to watch.

Remaking The Simpsons: Hit and Run in Unreal Engine 5

Not only do the videos highlight the power of the Unreal Engine. They also demonstrate how far game development has evolved over the past twenty years.

I find these types of personal projects inspiring and motivating. Helping fuel my passion for game development. I have always avoided Unreal Engine in favour of other (simpler) engines, such as Godot and GameMaker. However, these videos do open my eyes to what is possible with Unreal Engine as a single developer.

I look forward to seeing how the project evolves and would love to see some formal acknowledgement/backing from Activision Blizzard.

Finally, I recommend you subscribe to reubs YouTube channel, as well as support him via his Patreon.

Implementing Zero Trust

Sat, 12 Nov 2022 00:00:00 -0600

Zero Trust was first referenced in 1994. However, it was not until 2001 that the modern definition of Zero Trust started to emerge. In 2023, Zero Trust is a common “buzz term” across cybersecurity, frequently referenced by vendors, researchers, analysts, etc.

I am an advocate of Zero Trust, superseding the now inadequate moat/castle strategy. Over the years, I have written a few articles about Zero Trust, as well as partnered with Microsoft to promote the core principles and concepts.

The terminology “Zero Trust” can be confusing. Therefore, when asked, I usually respond with the following definition.

Zero Trust is a strategy for IT security, which assumes that internal and external threats always exist and that all networks are inherently hostile.

The traditional concept of a perimeter is disregarded, shifting the emphasis to Identity Access Management (IAM), where all connections accessing business assets must first be authenticated and authorised.

In this scenario, IT services are individually secured and monitored at source, following the Principle of Least Privilege (PoLP).

Therefore, a holistic security posture can be maintained regardless of the user or IT service location, with appropriate threat identification and blast radius controls.

The challenge with this definition (and Zero trust in general) is how it can be tangibly applied within a business, specifically an enterprise business.

Therefore, this article will aim to provide more context to Zero Trust, highlighting how I position and apply the core principles and concepts.

The diagram below is a high-level architecture overview, outlining the layers that contribute to the security model.

NOTE: This high-level architecture overview will not apply to every business. It is my perspective, which can be used as a guide (starting point).

From left to right, the diagram starts with the user (e.g., employee, contractor, customer) and proceeds through layers of security to access the desired runtime system (application), supported by a range of back office services (hosting, backup, asset inventory, etc.)

Vertically, the diagram highlights the need for holistic Governance, Risk and Compliance (GRC), as well as unified Security Operations, ideally fed via a standardised information/event management architecture.

The bullets below provide a short description for each layer, which in a perfect world would be enabled through programmatically defined automation.

This approach to automation allows policy directives and standards to be enforced proactively and consistently as controls, managed via code that can be easily verified to ensure compliance.

The bullets below describe each layer.

Edge Management: In the context of client-side (non-web) applications/data, specific capabilities to discover, manage and protect the user endpoint or the specific user endpoint applications and data. For example, Endpoint Management and/or Mobile Application Management (MAM).
API Management: In the context of modern web-first applications/data, specific capabilities to document, discover, access, manage, orchestrate, scale and secure API endpoints.
Security in Transit: Encryption in transit for modern web-first applications/data, alongside secure remote access capabilities to interact with services that are not web-first or Internet-facing. For example, Secure Access Service Edge (SASE).
Modern Authentication: Specific capabilities to prove an assertion, such as the identity of a user or computer system, including Identity Governance Lifecycle, Multi-Factor Authentication and Public Key Infrastructure.
Authorisation: Specific capabilities to manage access rights/privileges to services, with an emphasis on Least Privileged Access (specifically Privileged Access), supported by dynamic Conditional Access.
Network Security: Network-based capabilities with application-layer (Layer 7) inspection and management, supporting DNS/DHCP, as well as macro and micro-segmentation of secure zones. Segmentation is especially valuable for environments that must meet specific availability and/or compliance requirements.
Hardened Source Code: A case base tracked in version control, deployed via a standard pipeline that includes controls to proactively enforce quality, privacy, and security. For example, code, dependency, secret analysis, etc.
Runtime System: Real-time monitoring of the runtime environment to verify the running code and dependencies are up to date and operating as designed (no configuration drift, malicious code, etc.)
Backing Services: Connected resources, accessed via a URL or other locator/credentials stored in a separate vault. Backing services could include data stores, caches, messaging services, etc. Supporting capabilities cover data catalogue, labelling, monitoring, loss prevention, etc.
Security at Rest: Encryption at rest, alongside physical security controls.
Hardened Hosting: Capabilities to monitor and protect the hosting environments (Cloud, Data Centre, Edge), ensuring the foundations are hardened following the pre-defined documented standards.
Immutable Data Backup: Data protection capabilities to support backup and recovery, including an immutable filesystem to protect against ransomware attacks.
Asset Inventory: A dynamically maintained list of all IT assets (endpoints, appliances, services, applications, data), including the relevant metadata to manage responsibility assignment and dependency mapping, highlighting critical assets to enable risks-based decisions.

With the layers defined, a capability map can be defined (as highlighted in the diagram below). As previously stated, the required capabilities will depend on the specific business and risk posture.

That’s it! Hopefully, this information has provided more context regarding Zero Trust, specifically how it can be tangibly applied within a business. As an additional reference, I would recommend reviewing the The National Cyber Security Centre (NCSC) Zero Trust Architecture, which outlines eight key principles.

My Setup (Q4 2022)

Mon, 07 Nov 2022 00:00:00 -0600

Update - Hardware specification updated to reflect a new Apple MacBook Pro 16-inch (JAN-2023) and Framework Laptop 13 DIY Edition 13th Generation Intel Core.

Final Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

I switch between four devices (two desktops and two notebooks), which are primarily used for productivity, collaboration, software development, photo editing, video editing, virtual labs, gaming and game development.

My daily driver is a custom-built desktop PC. The full specification can be found below:

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
NVIDIA GeForce RTX 3090 FE 24GB GDDR6X
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

I also have an Apple Mac Studio (MAR-2022), which has two user profiles, one for me and the other for my wife. She is Head of Marketing at Winchester Science Centre, therefore is a frequent user of Serif Affinity Photo, Apple Final Cut Pro, etc.

Apple Mac Studio (MAR-2022)
Apple M1 Max (10-core - 8 Performance / 2 Efficiency)
32-core GPU (10.4 Teraflops)
16-core Neural Engine
64GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
4x Thunderbolt 4 (40Gb/s), 2x USB-C (10Gb/s), 2x USB-A (5Gb/s), 10Gb Ethernet, HDMI, 3.5mm Headphone, SDXC Card Reader

When remote or travelling, my daily driver is an Apple MacBook Pro 16-inch (JAN-2023). The full specification can be found below:

Apple MacBook Pro 16-inch (JAN-2023)
Apple M2 Max (12-core - 8 Performance / 4 Efficiency)
38-core GPU (13.6 Teraflops)
16-core Neural Engine
96GB Unified Memory (400GB/s Memory Bandwidth)
2TB SSD (7.4GB/s Read)
3x Thunderbolt 4, HDMI 2.1, 3.5mm Headphone, SDXC Card Reader, MagSafe 3
16.2-inch Liquid Retina XDR Display (3456x2234 @ 120Hz, 1600nits)

Finally, I use a Framework Laptop running (Fedora) for Linux-specific workloads (e.g., Docker, Ethical Hacking, etc.) I am a believer in the “right to repair”, therefore eager to support companies promoting this cause.

The full specification can be found below:

Framework Laptop DIY Edition
Intel i7-1370P 5,2GHz (14C/20T)
64GB Crucial DDR4 PC4-25600C22 3200MHz RAM
1TB Western Digital Black SN850X NVMe (7.3GB/s Read)
Intel Iris Xe Graphics
13.5-inch IPS LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

At home, the desktop PC and Apple Mac Studio connect to a 49-inch Super Ultra-Wide monitor, specifically the Samsung C49RG90.

The full specification of the monitor can be found below.

Model: Samsung C49RG90
Panel Size: 49-inch
Panel Type: VA
Panel Curvature: 1800R
Aspect Ratio: 32:9
Resolution: 5120x1440
Refresh Rate: 120Hz
Variable Refresh Rate: AMD FreeSync 2 (48-120Hz Range)
Response Time: 4ms (GTG)
Colour Accuracy: 125% sRGB, 92% Adobe RGB, 95% DCI-P3
High Dynamic Range: HDR1000
Brightness: 600cd/m2 (Typical), 1000cd/m2 (Peak)

My workspace is kept fairly minimal, thanks to the exceptional Secret Lab Magnus Pro XL sit-to-stand desk.

The photo below provides a closer look at the Samsung C49RG90 monitor.

Additional peripherals include speakers (AudioEngine A1), headphones (Razer BlackShark V2 and Bang & Olufsen BeoPlay H8), controller (Microsoft Xbox Elite Series 2), flight controller (Saitek X-55), steering wheel (Logitech G29) and virtual reality headset (Meta Quest Pro).

As highlighted in the photo below, my desktop PC and Apple Mac Studio sit under the desk, with the PC case door exposed providing easy access to the components.

Accompanying the desktop PC and monitor is an Original Prusa MINI+, which is an open-source 3D printer.

Finally, I use a Herman Miller Mirra 2 chair, which balances comfort and personalised ergonomics.

Home Server

Alongside my desktops and notebooks, I have a small home server, which is connected to my Samsung S95B OLED TV (65-inch). The server manages local and cloud backups, media streaming, and collaboration (video conferencing).

I selected an Apple Mac mini (NOV-2018) for the server, thanks to its excellent power efficiency (150W max), high-performance I/O (4x Thunderbolt 3, 2x USB-A 3.0) and small form factor. The Mac mini also offers versatile video conferencing capabilities covering Apple FaceTime, Zoom, Microsoft Teams, Google Meet, Cisco WebEx, WhatsApp Video, etc.

The full specification of the Mac mini can be found below:

Apple Mac mini (NOV-2018)
Intel Core i5-8500B 3.0GHz Base / 4.1GHz Boost (6C/6T)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
256GB PCI-e SSD
Intel UHD Graphics 630
2x 500GB Samsung Portable T5 SSD (USB 3.1 Gen2)

The 1256GB of local SSD storage is split across multiple drives for resilience, with automated backups being completed by Carbon Copy Cloner and Cloud Storage via Microsoft OneDrive.

To support my minimal (hidden-wire) entertainment setup, all of my AV equipment (Sony STR-DN1050 AV Receiver, Sony PlayStation 5, Apple Mac mini, Logitech Harmony Hub) is located in a cupboard under the stairs, with the required cables fed through the wall to the TV. This can be a little inconvenient when looking to insert/change physical media, but thankfully the cupboard offers plenty of room for ventilation.

Meta Quest Pro

Tue, 25 Oct 2022 00:00:00 -0500

Earlier today I received the Meta Quest Pro, which is the latest Virtual Reality headset from Meta (formally Facebook/Oculus).

My first experience of Virtual Reality was in the early-90’s when I stumbled upon a demonstration of a Virtuality 1000CS. However, I didn’t start to take Virtual Reality seriously until around 2016, with the release of the Oculus Rift Consumer Version (CV).

Over the years, I have owned multiple consumer Virtual Reality headsets, starting with Oculus Rift Developer Kit 2 (DK2) and most recently the Meta Quest 2.

I have also explored other Mixed Reality headsets/glasses, such as Google Glass and Microsoft HoloLens 2.

The Meta Quest Pro does not replace existing the Meta Quest 2, instead targeting a different audience. It shares the same core architecture (Qualcomm Snapdragon XR2), with the focus being new optics, sensors and controllers.

The specification can be found below.

Qualcomm Snapdragon XR2+
Adreno 650 Graphics
Pancake Lens and Optics Technology (Non-Fresnel)
1800x1920 (Per Eye) at 72/90Hz
12GB LPDDR5 Memory
256GB Storage
10x Cameras (5x External, 5x Internal)
Touch Pro Controllers (Qualcomm Snapdragon 662, 3x Cameras)

The Meta Quest Pro is £1499, which is a significant price increase from the £399 price of the Meta Quest 2.

To justify this price change, Meta describes the Meta Quest Pro as a business device, not targeting consumers. With this positioning, the Meta Quest Pro sits alongside other business-centric Mixed Reality headsets, such as the Microsoft Holo Lens 2 ($3,500) and Magic Leap 2 ($3,299).

I suspect this positioning will confuse and disappoint many Meta (and previous Oculus) customers that likely hoped to see a successor to the popular (and price-competitive) Meta Quest 2.

This was apparent in the recent Meta Connect 2022 Unscripted Talk from John Carmack, who believes Meta should be prioritising a low-cost, mass-market virtual reality headset.

The Meta Quest Pro is an elegant headset, with excellent build quality using premium materials.

The headset itself is reasonably heavy at 722g, compared to the Meta Quest 2 at 503g. However, the weight has been redistributed between the front (optics and system on a chip) and back (battery).

In my opinion, this design makes a big difference and I certainly prefer the Meta Quest Pro over the Meta Quest 2 for prolonged usage (even with the Elite Strap. With that said, Virtual Reality headsets are highly personal, similar to earphones, therefore individual comfort will vary.

The new pancake lenses reduce the size of the optical stack by 40% compared to the Meta Quest 2. They work by folding light inside the optical module, delivering 75% greater contrast with a 1.3x larger colour gamut.

The difference is immediately noticeable, positively impacting the design of the headset and image quality when used. For example, the Meta Quest Pro is not prone to lens flaring, which is a common complaint with existing Virtual Reality headsets.

Unlike the previous Touch Controllers, the Touch Pro Controllers are self-tracked, meaning that each controller has three cameras and a dedicated Qualcomm Snapdragon 662 SoC.

The removal of the sensor ring and the ability to track independently from the headset is a major positive. However, the extra horsepower forces a shift to rechargeable batteries, providing up to eight hours of battery life.

In use, the Meta Quest Pro feels exactly like the Meta Quest 2. Although the Meta Quest Pro targets business use cases, the software (operating system) is identical to the consumer range. In fact, there are no obvious enhancements that specifically target businesses. As a comparison, the Microsoft Holo Lens 2 supports a wide range of industrial standards, such as ISO 14644-14, ISO Class 5.0, UL Class I, and Division 2.

First-party applications such as Meta Horizon Worlds and Meta Horizon Workrooms are available for the Meta Quest Pro and do (in theory) support the new features, such as eye tracking.

However, even with billions of dollars invested by Meta, these applications continue to be very underwhelming. At this point, they feel like a rehash of ill-fated applications such as PlayStation Home, whilst being riddled with bugs, delivering an almost unusable (but sometimes comical) user experience.

Thankfully, there are some good third-party applications (designed for the Meta Quest and Oculus range of headsets), that all work as designed on the Meta Quest Pro.

Unfortunately, although the new features, such as eye tracking and the improved mixed reality mode can be generically enabled from the settings, individual applications require a specific software update to ensure full compatibility.

At the time of writing, there were very few third-party applications that demonstrated these features effectively. Therefore, it is difficult to determine their overall impact compared to other headsets that lack these features.

What I can state is that although the high-definition front-facing cameras certainly improve the Mixed Reality mode (compared to the Meta Quest 2), the quality is still unacceptable for everyday use.

For example, reading text in the “real world” through the Meta Quest Pro is very difficult, generally requiring the text size to be increased.

In many respects, the Meta Quest Pro feels confused, trying to appease the Virtual Reality community, whilst also competing against Mixed Reality headsets, such as the Holo Lens 2. This is most evident by the fact that the Meta Quest Pro is not fully enclosed, something you would expect from a dedicated Virtual Reality headset.

As a result of this design decision, there is a significant amount of light leakage from the sides and bottom. In theory, this is a negative for Virtual Reality, but a potential positive for Mixed Reality.

The Meta Quest Pro design even looks similar to the Holo Lens 2, as demonstrated by my son. Can you spot which headset is the Meta Quest Pro? It is not immediately obvious.

I expected to dislike these design choices, thinking they would break the immersion of Virtual Reality and potentially contribute to motion sickness.

However, surprisingly I did not find it to be an issue. After a few minutes of wearing the Meta Quest Pro, my focus recalibrated and essentially ignored the light leakage when using Virtual Reality applications.

Initial Impressions

Looking at the Meta Quest Pro headset in isolation (ignoring all other headsets and the price), my initial impressions are mostly positive.

The headset is well built and demonstrates potential, with advanced features such as pancake lenses, eye tracking, front-facing cameras, etc.

I do believe it suffers from attempting to bridge the gap between Virtual Reality and Mixed Reality, essentially compromising both modes. However, it still performs well enough to make it the best overall Virtual Reality headset I have used to date.

The video below shows my son playing Iron Man VR.

Unfortunately for Meta, the Meta Quest Pro is not the only headset on the market and the price is difficult to ignore when comparing features that most (consumer) users desire.

Therefore, if you are looking for a good quality consumer virtual reality headset for playing games, etc. You should purchase the Meta Quest 2 (£399) or PICO 4 (£379.99) (not the Meta Quest Pro). Alternatively, wait for the inevitable Meta Quest 3, which I would expect to be announced in 2023 and could even include more advanced features, such as a new System on a Chip and 120Hz Refresh Rate.

If you are a business, looking to explore Mixed Reality and the “Metaverse”, the Meta Quest Pro is an interesting option. However, consumer-centric software, poor quality first-party applications and limited business-centric features reduce its relevance, resulting in a questionable return on investment.

I am excited to see how the Meta Quest Pro evolves and where Meta goes next.

CircuitMess Nibble

Thu, 20 Oct 2022 00:00:00 -0500

Knowing my love for all things technology, my wife recently purchased me a CircuitMess Nibble for my birthday.

The Nibble is a small, retro gaming console that must be physically built and programmed (C++ and CircuitBlocks).

As highlighted by the articles below, my wife knows I enjoy completing STEM activities with my son (my daughter is still a little young), therefore the Nibble was a great gift!

The embedded video below highlights the Nibble in action.

The CircuitMess Nibble is very well packaged. However, the box only includes an EU charger. Therefore, if purchased for the UK market, CircuitMess provide a separate adapter, which is free but not included in the box.

Inside the box, all components are neatly packaged, including the required AAA batteries for the gaming console, solder and soldering iron safety stand.

The instructions are also excellent, with step-by-step instructions, supported by large, clear fonts and a colourful, inviting theme.

The build instructions are also available online, at circuitmess.com/build.

My son (seven years old) has never soldered before. Thankfully, CircuitMess provide details instructions, including a video tutorial featuring Collin Cunningham.

Some of the components are quite small and require force to install. Therefore, adult support throughout the build is advisable.

The risk of breaking the device is low, except for the screen, which includes eight narrowly spaced pins. As the LCD is also fragile, I would recommend helping any children with this part.

I have completed a range of software projects with my son. However, this is his first “real” hardware build (outside of Lego). He appeared to enjoy the process and I feel seven years old is probably the right age.

The only issue we had was the occasional bad solder joint and/or bridged solder joint. These can be easily rectified by re-heating the joint and cleaning the iron using the included sponge.

Overall, my son and I enjoyed the build process. The next step is to programme the device and load some games!

Day in the Life

Wed, 19 Oct 2022 00:00:00 -0500

I have been the global Chief Technology Officer (CTO) at an enterprise business since 2018.

Earlier this year, my role was expanded to become the combined global Chief Technology Officer (CTO) and Chief Information Security Officer (CISO).

With approximately six months in the role, I thought I would provide an update regarding the day in the life as a combined CTO / CISO.

Morning

My day starts between 05:30 and 06:30. I aim to wake naturally, without an alarm. As a morning person, this is rarely an issue.

I usually spend the first 30mins of my day reviewing any updates that occurred overnight. This includes direct messages, as well as news and social channels, with a focus on World Events, Information Technology, Cybersecurity, Business, and Sports.

Once I have myself ready, I reset the house for the day ahead. This includes tidying anything left out from the night before and preparing my kid’s school bags.

My kids wake at around 07:00. My son (seven years old) is a morning person (like me) and is fairly self-sufficient. My daughter (four years old) usually requires more “encouragement”, including help getting ready for school (e.g., clothes, teeth, etc.)

I prepare breakfast for my kids and make myself a PhD Synergy All-In-One Protein shake (247kcal, 38g Protein, 15g Carb, 5mg Creatine) and take a Centrum Advance Multivitamin. I rarely eat breakfast, as time is fairly limited.

With breakfast complete, my kids have approximately 45mins to complete any homework (e.g., reading, spelling) and play before school. If possible, I use this time to complete any “quick wins”, which could be personal or work-related. This includes a review of my emails, targeting anything that can be instantly actioned.

At 08:20, I drop my kids at school, before heading directly to the office or back home (depending on the day). I usually work from the office on Tuesdays and Thursdays, with a focus on in-person meetings and collaboration.

The Working Day

My work day is usually back-to-back, with a combination of meetings and focused time.

Throughout my career as an IT professional, I have worked in operations as an analyst and delivery as an engineer, across Infrastructure, Global Services, Commercial, and Research & Development (R&D).

In these roles, the majority of my time was spent as an individual contributor, supporting requirements as part of a wider team, therefore focused time was the priority (getting stuff done).

As I progressed into leadership roles, either as an architect or a director, the balance shifted towards meetings. Defining and communicating strategy, steering, influencing, coaching, etc.

My team are predominately based in India, Poland, UK and the US (Eastern Time). As a result, I usually connect with the India/Poland/UK team in the morning, with the afternoon dominated by the US.

The global nature of the team can make time zone alignment a challenge (+5:30 to -5:00). Therefore, I accept the traditional 9-to-5 hours are not viable, instead I structure my week knowing that the start and finish times will vary, whilst ensuring time is protected to promote a healthy work / life balance.

For example, I will usually find 50mins during the day to exercise (usually over the UK lunchtime). I am lucky enough to have a gym at my house, where I focus on stretching and weights (compound movements) or a metcon workout. I may also go for a run (or a ruck run), depending on how much time I have available.

Where possible, I combine my exercise with an opportunity to listen to a podcast (usually technology related) or listen / watch a pre-recorded sprint demo. The goal is to maximise the working hours, whilst still breaking away from my desk and ensuring I achieve some physical activity each day.

I find being able to exercise in the middle of the day very refreshing, reducing stress and allowing me to hit the afternoon with renewed energy and motivation.

Regarding Lunch, I eat at my desk while working (not ideal, but a necessary evil). I usually eat a MuscleFood Prepped Pot, which tastes great, can be prepared very quickly (4-7mins heating time), as well as relatively healthy and cost-effective (< £3.59 per pot). For example, the Caribana Chicken Pot (358kcal, 34g Protein, 33g Carbs).

I do not drink coffee, therefore I am fueled on tea and the occasional PhD Nutrition Diet Whey Protein shake (91kcal, 17g Protein, 2.8g Carb). I won’t eat much outside of lunch/dinner, usually some fruit and a cereal bar (e.g., Nature Valley) if I am feeling drained.

To help provide insight into the structure of my day, I have analysed my schedule over one month. The results can be found below, highlighting where I spend my time.

I have split my work across twenty-one categories, with a few examples provided for each (not exhaustive).

On average, I work 178h per month, which is approximately 119% of my contracted time. The fact that this is above 100% is not ideal, but not unexpected or unmanageable considering my role. My company and industry are very relationship-based, therefore I do have a lot of meetings, up to 8 per day, 160 per month (many of which are 30mins in length).

Knowing this dataset is an average, captured over a single month, I acknowledge the balance of these categories and time consumed varies, influenced by many variables (e.g., Priorities, Workshops, Personal Commitments, etc.)

Looking at the data in more detail, my top five categories are:

Executive Leadership (12%)
Team and People Development (11%)
Technical Leadership / Consultancy (8%)
Business Development and Innovation (8%)
Information Security Programme (7%)

However, it should be noted that specific themes, such as Architecture / Engineering and Information Security are represented across multiple categories. For example:

Combined Architecture / Engineering categories consume a total of 19% of my time, covering Enterprise, Solution, Data and Domain architecture and engineering.

Combined Information Security categories consume a total of 16% of my time, covering Governance, Risk and Compliance, Architecture and Operations.

Finally, I look for opportunities each day to step away from my desk. For example, if I have a scheduled 1:1 (in-person or virtual) I will check to see if we (or I) can walk + talk. If I am working from home, I also look for small windows of opportunity to complete some house chores (load/empty the dishwasher, washing machine, etc.) These are small tasks that left unchecked can contribute to stress.

Evening

Except for Monday, when I collect my kids from their afterschool club and take my son to football, I usually finish work at around 17:00. I ensure that 17:00 to 19:30 is protected in my calendar, spent with my family having dinner and usually playing a game before the bedtime routine (homework, bath, stories, etc.)

My wife also has a hectic work schedule as Head of Marketing at the Winchster Science Centre, therefore I am fortunate that she can coordinate after-school clubs (e.g., swimming, etc.) and prepares the evening meal (we subscribe to Gousto).

With the kids in bed (hopefully asleep), I occasionally need to jump back on another work meeting or if I am training for a specific event (e.g., Ultra Event, Marathon, Obstacle Course Race, Hyrox, etc.) I squeeze in an additional exercise session.

As a general rule, I try to avoid using my smartphone from 20:00 onwards, removing the temptation for work or unintentional “doomscrolling”. The only exception is a personal emergency or work major incident.

I wear Fitbit Charge 5 lifestyle tracker and average 7000 steps (I would like that to be more) and 55 Active Zone Minutes per day. Considering I have a “desk job” I feel these numbers are respectable.

To wind down before bed, my wife and I usually watch TV. I aim to be in bed around 22:30. I occasionally read before bed, although I do not always have the energy.

Conclusion

That’s it! This structure represents a fairly accurate view of my working week, with weekends dedicated to family and friends.

My wife and I do not have family that live close, therefore with two demanding jobs and two young kids, the working week is fairly intense (which can become a touch stressful).

However, with a supportive wife, happy kids and great friends. Combined with a family-orientated employer, I feel incredibly fortunate and thankful.

New Home Office

Mon, 17 Oct 2022 00:00:00 -0500

Earlier this month, I upgraded my home office.

This was my first major office upgrade in over six years, prompted by the release of the Secret Lab Magnus Pro XL sit-to-stand desk.

Over the past decade, I have been fortunate enough to have the flexibility to work from home, something that has become increasingly common post-pandemic, supported by the acceptance of hybrid working.

My setup requirements have not changed much over the years, with the desk and monitor being the focal point of the room. I use computers for a wide range of tasks, specifically productivity, collaboration, software development, photo editing, video editing, virtual labs, gaming and game development. Therefore, although I have some gaming-centric peripherals, I try to keep my setup fairly neutral.

I have experimented with different monitor setups, ranging from a single Apple Cinema Display (2010), three Eyefinity configured monitors (2014), and dual 4K 27-inch monitors (2017), resulting in my current setup, a single 49-inch Super Ultrawide.

The photo below highlights my outgoing setup.

One of my goals was to switch from a corner desk to a straight desk, reducing the overall surface area, and providing more floor space in the office itself. I also hope to reorientate my seating position, as the window facing the monitor can cause distracting glare and reflections.

The Secret Lab Magnus Pro XL

The Secret Lab Magnus Pro XL sit-to-stand desk is a sight to behold.

At 177cm in length and 80cm wide, it is a serious desk, with a lot of surface area (although less overall surface area when compared to my previous corner desk).

The all-metal construction is incredibly robust, supporting up to 120kg in weight (twice my body weight) and millimetre-specific height adjustment from 650mm to 1250mm.

Unlike other sit-to-stand desks, the control panel is fully integrated into the desk itself, providing a clean, premium aesthetic. It includes touch-sensitive buttons for manual height control, four customisable pre-sets and a physical lock switch.

The sit-to-stand motor is also fully-integrated, hidden inside the legs and connected via a single electrical socket on the bottom left leg. This electrical socket not only powers the desk but also connects to a hidden cable that provides an additional electrical socket directly to the large cable management tray.

As a result, the cable management tray, which extends the full length of the desk, conceals all cables, with enough room for a large power extension lead. The cable management tray is accessible via an easy-access rear-hinged cover, which provides enough room to route (hide) cables and mount a monitor arm.

Finally, the desk complies with all relevant safety standards and regulatory requirements, including the use of a thermal circuit breaker, insulated 12AWG cables, segmentation and grounding, as well as anti-collision detection.

In summary, the Secret Lab Magnus Pro XL sit-to-stand desk is the “Rolls Royce” of home office desks, with no-expense-spared materials, build quality and features.

Regarding price, it is certainly a premium product, at £829. However, considering the feature set, it compares well against the premium competition.

The Build

I was incredibly impressed by the packaging and documentation of the Secret Lab Magnus Pro XL. Every part was well protected and structured, making the “out of the box” experience a pleasure. My only minor feedback relates to the amount of plastic used to wrap the parts, which could arguably be reduced or replaced by more environmentally friendly materials.

I assembled the desk alone in about 30mins. However, I would recommend at least two people. The desk is heavy and must be flipped as part of the build process.

All required tools and fixings are provided, with each part having a specific installation location and orientation. As a result, it is almost impossible to go wrong.

The team over at ShortCircuit recently reviewed the desk, including the build process. I highly recommend watching their video for more details.

I also purchased the Secret Lab cable management bundle and Secret Lab MAGRGB XL Diffused RGB Strip.

Both products connect via strong magnets, which makes installation and adjustments very simple and highly intuitive.

Conclusion

Overall, I have been incredibly impressed by the Secret Lab Magnus Pro XL sit-to-stand desk, which sets the tone of the room (you are here to work).

Although the desk is the most exciting and impactful change, I did take the opportunity to decorate the room and replace some peripherals/furniture.

Most notably, I purchased an Ergotron HX Desk Monitor Arm with HD Pivot, which is one of the few on the market that can hold a 49-inch Super Ultrawide monitor, like my Samsung C49RG90.

Unsurprisingly, the monitor arm is heavy-duty, with a hinge that can support up to 19.1kg (the Samsung C49RG90 is 11.6kg). It is fully compatible with the Secret Lab Magnus Pro XL, installed using the included clamp.

Unfortunately, the use of premium materials results in a high price of £269.99.

I replaced my previous three-bulb ceiling light with a Phillips Hue Aurelle Square Panel Light. This diffused light reduces the glare on the display and also provides “warm” colour options, ideal for a range of different working scenarios (video conferencing, evening working, etc.)

In addition, I installed a Navaris Magnetic Glass Whiteboard on my door, which faces the desk. This doesn’t sound exciting, but when combined with the Insta360 Link 4k webcam, it can be used as an augmented whiteboard whilst video conferencing.

Finally, I continue to use a Herman Miller Mirra 2 chair, which aims to balance comfort and personalised ergonomics in a single design.

The photo below highlights my new setup, including art work from my wife (Whaam!, Roy Lichtenstein).

The desk itself is essentially wall-to-wall but provides a lot of space thanks to the monitor arm.

That’s it! My new home office setup is complete, hopefully serving me well for the years to come.

GitHub Success Story

Fri, 23 Sep 2022 00:00:00 -0500

Earlier today, GitHub published a success story highlighting the great work completed by my DevSecOps team to standardise and automate the developer experience, whilst enforcing security standards and enhancing security visibility.

With digital and data business models becoming the standard across every industry, the ability to ship reliable, scalable and secure code is more critical than ever before.

As a result, the team seized the opportunity to position GitHub as the common DevSecOps platform across the enterprise, including pre-built patterns (using HashiCorp Terraform) that unlock speed to value and agility through reuse, whilst promoting concepts such as innersourcing to drive continuous improvement.

These patterns, combined with GitHub Actions and GitHub Advanced Security, ensure security, privacy and quality standards are proactively enforced through programmatically defined controls.

The result, in 2021, infrastructure and application provisioning took an average of 600 hours (> 25 days). Through the new DevSecOps automation foundation, this lead time was reduced to 30 mins (< 1 day).

I highly recommend reviewing the success story for more details and/or connecting with the project lead, Calum Bell.

3D Printed Framework

Fri, 19 Aug 2022 00:00:00 -0500

Since May, I have been using a Framework Laptop as my daily driver.

I have written a few articles highlighting my experience.

I recently upgraded my laptop, specifically the Intel 12th Generation mainboard, adding 10 Cores and 12 Threads.

This left me with my old Intel 11th Generation mainboard, which includes the very capable Intel i7-1185G7.

Therefore, I decided to 3D print the Framework PC (mainboard) case with VESA mount, transforming the Framework Laptop into a Small Form Factor (SFF) PC.

Although this would not be my most complex print, it would be my longest, with six files:

I would be using my Original Prusa MINI+, with an estimated print time of 27 hours.

The case is designed to be modular, meaning it can be printed on 3D Printers with a smaller hotplate.

I printed the parts over several days, starting with the bottom left.

With all parts complete, the case can be assembled (no tools required).

The build precision tolerance is not perfect, but certainly good enough when acting as an Small Form Factor PC, with the appropriate gaps for cooling, expansion cards and Wi-Fi antenna points.

The middle of the case is removable, providing convenient access to the mainboard.

The case includes the required mounting holes for the mainboard, compatible with the original screws. However, screwing into PLA can be imprecise.

This fun little experiment highlights the potential when the hardware is designed to be serviceable. With very little effort and almost no additional cost, I now have a Small Form Factor PC, only missing memory, storage and a Wi-Fi module.

I am excited to see what the community come up with next!

Framework Laptop - Upgrade

Thu, 18 Aug 2022 00:00:00 -0500

Since May, I have been using a Framework Laptop as my daily driver, running Fedora 36.

I have written a few articles highlighting my experience.

Fulfilling the vision of Framework, I recently ordered two upgrade parts:

The hinge targets a personal criticism highlighted in my previous article. The original hinge was not overly robust, resulting in noticeable display wobble when typing. Likely subjective, but this was a point of frustration for me.

The upgrade kit is the real star of the show, specifically the Intel i7-1280P. The new Top Cover is a welcome addition, but not something I would consider essential.

As highlighted in my previous articles, the original 11th Generation Intel i7 Processor (Tiger Lake) is very capable for daily usage (productivity and collaboration), but certainly not a high-performance part. The full specification can be found below.

Processor: Intel i7-1185G7
Cores: 4
Threads: 8
Max Turbo Frequency: 4.8GHz
Base Frequency: 3.00 GHz TDP-Up (1.2GHz TDP-Down)
Cache: 12MB
TDP Maximum: 28 W
TDP Minimum: 12 W
Graphics: Intel Iris Xe Graphics (1.35GHz)

The 12th Generation Intel Core i7 Processor (Alder Lake) is a significant improvement, with more cores, threads and cache, but also higher power consumption. The full specification can be found below.

Processor: Intel i7-1280P
Cores: 14 (6 Performance / 8 Efficiency)
Threads: 20
Max Turbo Frequency: 4.8GHz (3.6GHz Efficiency)
Base Frequency: 1.8GHz (1.3GHz Efficiency)
Cache: 24MB
TDP Maximum: 64 W
TDP Base: 28 W
TDP Minimum: 20 W
Graphics: Intel Iris Xe Graphics (1.45GHz)

I am not aware of any other laptop where you can upgrade the mainboard, incrementing the architecture generation (Intel 11th Generation to 12th Generation), whilst adding 10 Cores, 12 Threads and doubling the Cache (12MB to 24MB).

In my opinion, the launch of the Intel 12th Generation mainboard validates the “right to repair” model defined by Framework, as it is an order of magnitude more complex than “simple” upgrades, such as the display, battery, etc.

Upgrade Experience

I decided to replace the mainboard, display hinges and top cover at the same time, although this could be completed incrementally. All other components, including the memory and storage, were reused from my Intel 11th Generation laptop.

Although it is technically possible to swap the mainboard without reinstalling the operating system, I had already decided to do a clean install of Fedora 36 post upgrade.

The entire upgrade process is completed via one tool, the included screwdriver (T5/PH0 double-sided bit). Access to the mainboard is achieved via the input cover, which is released via 5 magnetic screws on the bottom cover

All primary components include a QR code, which links to the specific user servicing guide. This is a nice touch and very pro-consumer.

I started with the mainboard itself, which includes 34 steps, but I am pleased to report was very simple.

With the seven connectors and five screws removed, the mainboard easily lifts from the chassis.

Even with the more powerful processor, the cooling solution on the Intel 12th Generation mainboard is similar in design, size and weight to the Intel 11th Generation. It will be interesting to see how effectively it cools the additional 10 Cores and 12 Threads.

The mainboard upgrade took less than 10 minutes to complete. The only area that requires additional focus was the removal of the connectors, as some of the flex cables are delicate.

Next, was the hinge and top case upgrade.

This was a little more fiddly. The top case upgrade includes 21 steps, requiring the removal of the display, camera module and Wi-Fi antenna cables, which are all delicate components.

I had to pay special attention to the Wi-Fi antenna cables, which include tiny connectors and are routed across the mainboard into the top case. This is not a criticism of Framework, simply a required compromise to ensure industry standards are maintained.

The hinge (two parts) was a simpler 18 step process, with the only minor frustration being the strength of the magnets used as part of the Framework top case design, which tended to attract the screws from the screwdriver when installing.

If you plan to upgrade the top case and hinges, I would recommend purchasing them together, knowing the installation requires similar steps.

With the new hinges and top case installed, I could immediately feel the difference, specifically, the hinges, which require more force to manoeuvre, removing the wobble when typing.

The top case impact is less apparent, which considering it was the most “complex” part to upgrade, makes me question the value.

The hinges and top case took around 20 minutes to upgrade, with most of the time spent ensuring the display, camera module and Wi-Fi antenna were removed and reinstalled without damage.

With the upgrades complete and existing memory and storage were re-installed.

The UEFI BIOS confirmed all the components were detected as expected and the new Intel 12th Generation mainboard was running the latest BIOS version.

At which point, I inserted the Fedora Installation Media (USB Drive) and reinstalled the operating system.

Upon completion, I am pleased to report the drivers for the Intel 12th Generation mainboard were immediately detected and initialised by Fedora 36 (Linux Kernel 5.18.17).

The only issue I have encountered relates to the display brightness hotkeys, which failed to respond. Thankfully, the Framework team were aware of the issue and had posted a knowledge article.

In short, there is a conflict between the Ambient Light Sensor (ALS) and the display brightness hotkeys. Therefore, at this time, only one can work. I always disable the Ambient Light Sensor on Linux (I find it to be unreliable), therefore decided to enable the hotkeys using the following command. Once executed, you must restart your laptop.

sudo grubby --update-kernel=ALL --args="module_blacklist=hid_sensor_hub"

That’s it! In 30 minutes, I have added 10 Cores, 12 Threads, a more rigid hinge and a robust CNC top case.

Great work to the Framework team for delivering against their “right to repair” promise and producing a high-quality product with excellent (user-centric) documentation.

Performance

Regarding performance, I used my previous Geekbench score and Docker build time as a reference.

Intel i7-1185G7

Single-Core Score: 1727
Multi-Core Score: 6134

Intel i7-1280P

Single-Core Score: 1860
Multi-Core Score: 10822

As you can see, Geekbench highlights a healthy 76% increase for multi-core workloads.

This result is not a surprise, considering the Intel 12th Generation mainboard includes 10 additional cores and 12 additional threads. The single-core performance is less exciting, showing a modest 7.7% increase.

Regarding Docker, I executed my standard build, a project that includes over 4000 individual files and 800MB of data.

Intel i7-1185G7: 55 Seconds
Intel i7-1280P: 38 Seconds

This was an impressive result, highlighting a 40% decrease in build time. Any build time decrease is always welcomed, as it can be very disruptive when integrating code throughout the day.

Finally, it is worth mentioning battery life.

Actual usage is very respectable for a Linux laptop, but cannot rival a modern Apple MacBook. With a varied workload, I have been achieving between 6 to 8 hours, although more consistently able to reach 8 hours when compared to the Intel 11th Generation.

The main issue with the Intel 11th Generation was the battery drain when sleeping. As a result, I was forced to switch the sleep state from the default “s2idle” to “deep”.

Previously, with the sleep state set to “s2idle”, the battery percentage would drop by approximately 15% per hour, resulting in a maximum sleep time of 6.5 hours before the battery would die (very frustrating).

With the sleep state set to “deep”, the battery percentage drop improved to around 1.6% per hour, achieving a more respectable (but still bad) 14.4% drop overnight (nine hours). However, this change has a trade-off regarding wake time, which can take upwards of 15 seconds.

The Intel 12th Generation appears to deliver better results, achieving an equivalent 1.6% per hour battery drain with the sleep state set to the default “s2idle”.

Compared to a modern Apple MacBook (or even a Windows laptop), these sleep numbers are still horrible. However, they are now at least manageable with the Intel 12th Generation and will hopefully continue to improve over time with new firmware/software releases.

In conclusion, I am very pleased with the upgrade process and improved performance, specifically multi-core and battery life when sleeping.

Things are still not perfect, but certainly, a positive step forward, building upon an already very solid foundation.

I look forward to seeing what comes next for Framework, maybe the addition of an AMD Ryzen-based mainboard and/or new display options offering anti-reflective coating, higher refresh rates, etc.

John Carmack

Fri, 12 Aug 2022 00:00:00 -0500

John Carmack, legendary programmer and co-founder of id Software recently completed an in-depth interview with Lex Fridman.

I consider Carmack one of the most influential thought leaders in technology, alongside Linus Torvalds, Steve Wozniak, Bill Gates, Tim Berners-Lee, etc.

In the context of programming, Carmack is notable as his contributions extend multiple decades, with most other technology leaders transitioning to a business and/or ambassador role.

Carmack is best known for being the lead programmer of many blockbuster video games including Commander Keen, Wolfenstein 3D, Doom, Quake, etc. He was also CTO of Oculus VR, delivering the demo that launched the modern era of virtual reality.

The Fridman interview is certainly long (five hours), but provides an incredible insight into the mind of Carmack, covering his career highlights, the state of programming, personal experiences, controversy, work/life balance, and future innovation across space exploration and artificial intelligence.

If you are interested in technology (specifically programming), I highly recommend you watch the interview.

In a world dominated by micro-blogging and brand-endorsed influencers, this type of detailed, unbiased, uninterrupted interview from a credible source is very rare.

Framework Laptop - Update

Sun, 07 Aug 2022 00:00:00 -0500

Since May, I have been using a Framework Laptop, running Fedora 36.

With three months of usage, I thought I would share an update regarding my experience.

Performance and Reliability

The specification of my Framework Laptop can be found below.

Framework Laptop DIY Edition
Intel i7-1185G7 4.80GHz (4C/8T)
64GB Crucial DDR4 PC4-25600C22 3200MHz RAM
1TB Western Digital Black SN850 NVMe (7GB/s Read)
Intel Iris Xe Graphics
13.5-inch LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

Considering its relatively low-powered processor (Intel i7-1185G7 - 28 W), the Framework Laptop has been impressively performant for daily productivity and collaboration tasks.

For example, my common workflow includes the following applications:

InSync (Google Drive and Microsoft OneDrive)
Google Chrome (Multiple Tabs)
LibreOffice
Krita
YOGA Image Optimizer
Microsoft Visual Studio Code
Docker
QEMU/KVM
OBS Studio
1Password
Microsoft Teams

All applications perform well (highly responsive), except for Microsoft Teams, which tends to “stutter” and severely impact battery life when the video is enabled.

It would be easy to assume this issue is unique to the Linux version of Microsoft Teams, but I have seen similar results with macOS and even Windows. Therefore, I have found accessing Microsoft Teams via the browser to be a better overall experience.

Regarding performance numbers, I did share a few benchmarks in my previous article. In short, all results (CPU, GPU, Memory, Storage) match industry expectations and I am pleased to report performance has remained consistent over time.

I have not seen any evidence of thermal throttling impacting real-world performance, even with an ambient temperature of 35+ degrees Celcius. This is impressive, considering the relatively small cooling solution.

Fedora itself runs amazingly well on the Framework Laptop! Outside of a few occasional application errors, the operating system has been “bulletproof”, with perfect hardware compatibility out of the box.

I rarely shut down the laptop, which over a week of work would usually highlight any software/hardware instability. To date, I have not experienced a full system crash.

Overall, I would state this is the best Linux laptop I have ever used/owned, including more expensive alternatives such as the Dell XPS 13, Dell XPS 17 and Lenovo ThinkPad X1 Extreme.

Build Quality

The Framework Laptop is built extremely well. The actual build quality (materials and manufacturing refinement) does not compete with premium options (Apple Mac) but it is certainly high quality.

I have used the laptop from multiple locations, including home, work and travelling.

I treat the laptop with respect, but it has certainly taken its share of “bumps and clunks”, especially when travelling, where it is frequently thrown into my bag without a protective case.

In addition, I have disassembled the laptop multiple times, demonstrating the serviceable features to friends and coworkers. This is clearly “above and beyond” what would be expected as standard.

I have owned other laptops that have shown signs of wear and tear very quickly, including scratches and marks across the palm rests and ports, as well as dents on the top cover.

I am pleased to report that this has not been the case with the Framework Laptop, which looks as good today as it did when purchased.

Unfortunately, not everything is perfect…

The standard display hinge is weak and tends to wobble when typing (even on a solid surface). The display lid is also more flexible than most, which doesn’t impact daily usage, but makes the laptop feel less “premium”.

I suspect this is a required compromise to ensure the display is serviceable, making it more difficult to reinforce the bezel or cover the display with protective glass.

Thankfully, highlighting the power and flexibility of the Framework philosophy, the team have responded to these areas of feedback, with a new 4kg display hinge and a more robust CNC top cover. I will likely purchase both in the future.

Finally, it is worth mentioning the expansion cards, which are a huge selling point but can be very difficult to release (requiring a lot of pressure).

Ironically, the locking mechanism appears to be very small (a potential point of failure), therefore it is a little surprising how well they secure the expansion cards.

Usability

Overall, I have enjoyed using the Framework Laptop. When assessed as a total package, including the price, serviceability, performance and build quality, I would argue it offers a compelling value proposition.

With that said, it is worth acknowledging the strengths/weaknesses of the individual parts:

Display: Great resolution and aspect ratio. Image quality is average. However, my main complaint is the reflective nature of the display, which makes it difficult to use in certain lighting.
Keyboard: Perfect! Although subjective, the Framework Laptop has one of the best laptop keyboards I have ever used.
Trackpad: Good, but not great. Certainly not comparable to an Apple MacBook, but equal to most Windows laptops I have used, with good sensitivity, responsiveness and reliability.
Fingerprint Reader: Acceptable: A useful addition that is not as convenient as Windows Hello or as accurate/responsive as Apple TouchID.
Camera/Microphone: Acceptable. The camera and microphone are perfectly adequate for basic video conferencing, although not something you would want to rely upon for anything more professional.
Speakers: Bad. Maybe I have been spoiled by the latest Apple MacBook Pro or there is a driver issue with Linux (I have not tested Windows natively), but the speakers simply sound bad. Be prepared to use headphones.

The good news, each part listed above is independently serviceable. Therefore, just as we have seen with the new hinge and top cover, Framework could conceivably produce and sell individual upgrades.

For example, I would love to see them deliver a new display with an anti-reflective coating or a new high-resolution camera module.

This is completely viable with the Framework Laptop, something that would be almost impossible with any other laptop.

Battery Life

The last area to highlight is battery life.

This is a controversial area, as there are known limitations with the Intel 11th Generation architecture and Linux. Therefore, please recognise that your experience will be very different if running Windows natively.

As stated in my article “Framework and Fedora”, I was forced to switch the sleep state from “s2idle” to “deep”. This change results in a slower resume time, but it dramatically improves the battery life when sleeping.

With that said, battery life is still fairly average.

When sleeping, the battery percentage dropped by approximately 1.6% per hour, meaning a total of 14.4% overnight (nine hours).

In use, avoiding sustained “heavy” workloads, I would average around 6 hours of battery life, which could easily drop to below 4 hours with more demanding (Encoding, Development Builds) or less well-optimised (Microsoft Teams) workloads.

Conclusion

Even with its faults, I love the Framework Laptop!

Yes, I am biased toward the mission of Framework to produce a laptop that respects the consumer’s right to repair and upgrade.

However, even when reviewing in isolation of the overall mission, the Framework Laptop is excellent, delivering a “best in class” Linux experience.

What is next for my Framework journey? I have ordered the following two upgrade parts from the Framework marketplace:

I am most excited to see how the Intel i7-1280P (14 Cores / 20 Threads) processor performs, recognising the significant increase in core count and power.

This upgrade will also unlock my existing Intel i7-1185G7 mainboard, which I plan to reuse with the 3D printed Framework PC (Mainboard) case.

Stay tuned for updates.

Cybersecurity Hardware Tools

Tue, 02 Aug 2022 00:00:00 -0500

Last month, I shared a list of safe and legal “Cybersecurity Resources”, covering websites, videos, training, ethical hacking, etc.

In the article, I also highlighted three hardware tools, specifically the Hak5 Rubber Ducky, Hak5 WiFi Pineapple and Flipper Zero.

This article will provide more details regarding these hardware tools and their usage.

NOTE: These tools must be used legally/ethically. For example, I use them for education, penetration testing, etc.

Hak5 Rubber Ducky

The Hak5 Rubber Ducky is a seemingly harmless USB drive (USB-A and USB-C interface). However, it is actually a hotplug attack tool, targeting keystroke injection. It is a hacker culture icon (as seen on Mr. Robot).

The premise, computers trust humans and humans use keyboards. Therefore, the Hak5 Rubber Ducky pretends to be a USB keyboard, automating keyboard entries.

Through the use of DuckySrcipt (scripting language), automated payloads can be easily created and executed. This includes features to fingerprint devices (operating system identification), and mimic human typing cadence, whilst evading endpoint restrictions and firewalls.

Hak5 WiFi Pineapple

The Hak5 WiFi Pineapple (Mark VII shown below) is a portable wifi penetration testing tool.

It can perform and automate a wide range of wifi-based campaigns, including auditing, surveillance, reconnaissance, man-in-the-middle, etc.

As highlighted by the video below, the Hak5 WiFi Pineapple includes an intuitive web interface for configuration.

Finally, it includes three dedicated role-based radios, with three high-gain antennas, supporting 2.4GHz 802.11 b/g/n and 5GHz 802.11 ac via a module.

Flipper Zero

The Flipper Zero is a standalone, multi-functional device developed for interaction with access control systems.

It is highly versatile and able to read, copy, or emulate RFID and NFC tags, radio remotes, infrared transmissions, iButton, and digital access keys. It also includes a GPIO interface and can act as a BadUSB device, targetting keystroke injection and accepting the previously mentioned scripting language (DuckyScript) used by the Hak5 Rubber Ducky.

The Flipper Zero firmware is based on the FreeRTOS operating system, with its own software abstraction over the hardware layer.

Like the Hak5 Rubber Ducky, it has become a hacker culture icon, being banned from Amazon and certain geographies.

Cybersecurity Dashboard

Sun, 31 Jul 2022 00:00:00 -0500

In May, my role was expanded to include Chief Information Security Officer (CISO).

A key responsibility is to represent cybersecurity as part of the Executive Committee and Board of Directors, covering the strategic direction, maturity, risks, incident management, awareness and education.

This has become increasingly important over the past year, with proposed rules from the SEC to enhance and standardise the disclosure of cybersecurity risk management, strategy, governance, and incident reporting by public companies.

To support this outcome, each quarter, I present a standard dashboard to help facilitate the discussion. In the spirit of transparency and collaboration, I thought I would share the dashboard template.

The dashboard includes eight sections, highlighting (and wherever possible, quantifying) key metrics associated with cybersecurity.

The sections were selected based on their relevance to the business. Therefore, if you plan to reuse the template, please modify/replace the sections as required.

NIST CSF Maturity Score:

The NIST Cybersecurity Framework (CSF) is a commonly used set of guidelines for mitigating organisational cybersecurity risks, based on established standards, guidelines, and practices.
The maturity assessment is scored against the 5 functions of the NIST Cybersecurity Framework (23 categories and 108 subcategories). Scoring is between 1 (No Control) and 5 (Enforced/Automated at Scale).
The industry score highlights the industry average score. If the industry average score is not known, the target score would be a good replacement.

Annual InfoSec Spend:

The total spend for Information Security includes direct (Infomation Security budget), indirect (other budgets) and people (FTE/Contractor).
The industry spend highlights the industry average, which for enterprise businesses is commonly between 7% and 10% of the IT budget (Ernst Young).

External Rating:

The external cybersecurity rating provides an external perspective from a reputable source, for example, Mastercard. Ideally, the score would incorporate software, infrastructure, geolocation, hosting providers, third parties, domains, systems and configuration.
External ratings are used to assess the material cybersecurity risk of a business before investments, partnerships, mergers, acquisitions and/or divestitures.
The industry score highlights the industry average score. If the industry average score is not known, the target score would be a good replacement.

Priorities:

The top five priorities for Information Security, ideally complimenting the business strategy and mitigating any enterprise and external risks.
Each priority includes a measurable outcome, ensuring progress can be tracked against a specific deliverable.

Assets:

The managed IT assets, covering the users (active identities), endpoints (Windows, macOS, iOS, Android), virtual machines (all hosting environments) and application services (SaaS, Serverless, Containers).
The managed percentage indicates the status of the assets, covering discovery, hardening, patching and advanced threat protection.

Awareness:

The results of key awareness and education initiatives. For example, a social engineering phishing campaign, highlighting the number of emails successfully “delivered”, “reported” and “clicked”.
The industry score highlights the industry average score. If the industry average score is not known, the target score would be a good replacement.

Detect & Respond:

The number of alerts detected, reviewed and investigated via the Security Operations Center (SOC), including the Median Time to Detect and Resolve.
The number of major incidents detected (hopefully zero). A major incident can be considered an unauthorized occurrence on or conducted through an information system that jeopardises the confidentiality, integrity, or availability of the information system or any information residing therein.

Risks:

The top three risk vectors (including the primary rationale) and the risk status of critical assets.

Each section includes a Red, Amber, Green (RAG) status, which can be used to help focus the discussion.

The iconography used as part of the “Priorities” and “Risks” sections can be updated as needed, with the colour (RAG) highlighting a sub-status.

I won’t claim this to be the perfect cybersecurity dashboard. However, I believe it does a good job of highlighting key information in a simple to consume format.

The most important thing is for the dashboard to facilitate a robust discussion with the Executive Committee and Board of Directors.

Windows 365

Sat, 30 Jul 2022 00:00:00 -0500

In my last article, I installed a Kernel-based Virtual Machine (KVM) on Fedora, enabling client-side (offline) virtualisation.

As previously highlighted, thanks to the use of browser-based applications and the growing support for Linux, 99% of my work can be completely natively from Linux. However, there are specific workloads that force me back to Windows.

Although client-side virtualisation is a good option, providing offline support and “local” performance, the need to install, configure and maintain a second operating system can be a pain.

As a result, I have also been testing Windows 365 from Microsoft, provisioned through my work.

In its simplest form, Windows 365 is a Virtual Desktop Infrastructure (VDI) service, similar to VMWare Horizon and Citrix DaaS.

It delivers a highly available, scalable and persistent (1:1) virtual machine (Cloud PC), accessible via any modern browser or standalone applications for Windows, macOS, iOS and Android.

Windows 365 has some unique advantages for enterprise businesses, specifically the unified user experience as part of Office 365 and full integration with Microsoft Endpoint Manager (AKA Microsoft Intune).

As with any VDI service, Windows 365 requires a persistent internet connection, with the quality of the experience being dependent on the available bandwidth and latency.

However, the primary barrier for most will be price, with a 4 vCPU, 16GB RAM, 256GB Storage plan costing £63.90 user/month (£766.80 user/year). As Windows 365 is reliant on the Azure network infrastructure, outbound (egress) charges are also applied, including Office 365 (OneDrive) data synchronisation and RDP network traffic.

When compared against laptop ownership with the equivalent specification, assuming a minimum three-year life cycle, Windows 365 is an expensive (£2300.40) proposition. However, as any enterprise customer of Microsoft will know, the list price is not indicative of the final price.

Windows 365 can be accessed via the web (windows365.microsoft.com or via Microsoft Remote Desktop.

Windows 365 Overview

Windows 365 can be launched from Office 365 for Web by clicking the “Apps” menu (top left). This helps Windows 365 feel like an integrated part of the Office 365 experience, following a usage pattern known by most users.

The Windows 365 Cloud PC selection web user interface is not well optimised, with a lot of unused screen real estate. I assume if you have access to multiple Cloud PC instances, the user interface would be more evenly populated.

However, at the current pricing and knowing that Windows 365 delivers a persistent (1:1) Cloud PC, I am not sure how many users (excluding administrators) would have access to multiple instances.

When a Cloud PC is launched, the session settings menu appears, providing quick access to important configuration items.

This is a useful prompt, allowing the user to enable the local Printer, Microphone and/or Clipboard. For example, if you plan to use Voice/Video capabilities (Microsoft Teams) via the Cloud PC, you must enable the Microphone.

The session settings menu is fairly intuitive, arguably better than what I have seen from VMWare and Citrix, which can often overcomplicate end-user settings with unnecessary technical jargon (Networking Settings, etc.)

Finally, you are prompted to verify your identity. For example, your Azure AD credentials.

A persistent Windows Client PC will then launch. If you are using the enterprise version, your administrator will have configured your default corporate settings. In my case, a standard Windows 10 desktop (Windows 10 is still the default in most enterprise businesses).

The Cloud PC toolbar (located at the top) provides access to key settings and keyboard shortcuts.

I have been testing Windows 365 on my Framework Laptop, running Fedora.

There is no native Linux client for Windows 365. Therefore, I used Google Chrome as my browser, but also tested Microsoft Edge (no noticeable difference).

So far, the Windows 365 experience has been positive. Everything works as designed and I can certainly see the benefits for enterprise businesses, especially if they are a Microsoft 365 and/or Microsoft Endpoint Management customer.

The performance of the Client PC has been good. Perfectly adequate for productivity and collaboration tasks, including the use of Voice/Video capabilities. However, “power users” will likely find the reduced graphics performance and increased latency a deal breaker.

As with any VDI-based service, some restrictions limit the viability for certain user groups. Specifically, the need for a persistent, performant, reliable internet connection.

If you are predominantly an office or home user, you will likely be fine. However, if you are a frequent traveller (planes, trains) I suspect the inability to use your Client PC offline will be a frustrating limitation, meaning you would be better served by a laptop running Windows and/or a local virtual machine.

Windows 365 is not a cost-effective option for my use case, as I only require access to Windows as an exception (not my primary desktop). Therefore, the monthly cost does not offer a compelling return on investment.

However, it does make sense for users that require persistent access to a managed/secured Windows desktop, where provisioning, managing and/or securing physical hardware is not viable. For example, users working on a service desk and/or third-party service providers.

KVM on Fedora

Wed, 27 Jul 2022 00:00:00 -0500

Over the past few months, my daily driver (personal and work) has been a Framework Laptop running Fedora.

The laptop is great and runs Fedora amazingly well (compatible, fast and reliable). Thanks to the use of browser-based applications and the growing support for Linux, 99% of my work can be completely natively from Linux.

However, there are still a few frustrating exceptions that force me back to Windows via a virtual machine.

I have a long history with VMWare, prodominately within the enterprise, supporting vSphere, NSX, Horizon, etc.

As a result, I have always favoured VMWare Workstation (Windows, Linux) and VMware Fusion (macOS) for client-side virtualisation.

However, with this new setup, I thought I would finally make the jump to an open-source Kernel-based Virtual Machine (KVM). Specifically, a combination of QEMU, Libvirt and Virt Manager.

Setup QEMU, Libvirt and Virt Manager

Confirm your processor has Intel-VT or AMD-V enabled.

grep -E --color '(vmx|svm)' /proc/cpuinfo

Ensure “VMX” or “SVM” are listed in the output.

Confirm the Kernel-based Virtual Machine (KVM) module is enabled.

lsmod | grep -i kvm

Ensure “kvm” is listed in the output. For example, “kvm_intel”.

Install the required virtualisation packages.

sudo dnf install -y qemu-kvm libvirt virt-install bridge-utils virt-manager

Specific packages include:

qemu-kvm: An emulator and virtualization package with hardware emulation.
libvirt: Configuration files required to run the libvirt daemon.
virtinst: Command line utilities for provisioning and modifying virtual machines.
Virt-install: Command line tool for creating virtual machines.
bridge-utils: A set of tools for creating and managing bridge devices.
virt-manager: Graphical user interface for managing virtual machines.

Finally, install the required virtualization modules using the command below.

sudo dnf install -y libvirt-devel virt-top libguestfs-tools guestfs-tools

That’s it! QEMU, Libvirt and Virt Manager are now installed.

To launch a virtual machine, you must first start the virtualization daemon.

sudo systemctl start libvirtd

Confirm the demon is running.

sudo systemctl status libvirtd

Ensure the status is “active (running)”. You can also check the CPU and Memory utilisation.

If you plan to run virtual machines frequently, it might be worth starting the service on boot. This can be done using the following command.

sudo systemctl enable libvirtd

Finally, launch the Virt Manager application to create a virtual machine.

Alternatively, virtual machines can be created via the command line tool, virt-install.

If you plan to run Windows 10/11 as a guest virtual machine, I recommend reviewing the performance guide on Medium by Leduccc, which includes a lot of great tips and tricks.

Winchester Science Centre

Wed, 20 Jul 2022 00:00:00 -0500

The Winchester Science Centre is an independent educational charity, that inspires children across the south of England (and worldwide online) to engage in Science, Technology, Engineering and Maths (STEM).

As Chief Technology and Information Security Officer (CTO/CISO) at a purpose-driven enterprise business, I am passionate about STEM and very aware of the challenges associated with diversity, inclusion and inequality.

Therefore, I am a long-time supporter (and frequent visitor) of the Winchester Science Centre.

Disclaimer, my wife is Head of Marketing at the Winchester Science Centre. However, we visited the Winchester Science Centre with our family many times before her employment. It was our love for the experience that prompted her to explore career opportunities.

The Winchester Science Centre provides a range of experiences for the entire family, including:

Sound, Hearing and Vibration: An open area packed with interactive exhibits. Highlights include Europe’s largest playable guitar, the sonic rocket, where you can explore environmental acoustics and my personal favourite, gravity run, which propels a ball through a vertical course.
Explorer:Space: A second open interactive area, focused on space and space exploration. Highlights include the ability to programme a miniature lunar rover and exhibits that demonstrate the impact of infrared light and gravity. There is also a soft play area for smaller children.
Bio:Space: An outside area, set in the South Downs National Park, providing a great place to relax and absorb nature, while the kids play on the mini adventure playground.
The Planetarium: The UK’s largest standalone planetarium, with a stunning 360-degree screen, with narrated shows that provide a unique insight into space, our solar system and beyond.

Alongside the open areas, interactive exhibits and Planetarium, the team run regular science shows at the Science Theature and explosive science demonstrations, known as Science Live.

For example, the photo below is my son exploring an exhibit that highlights the delicate balance between people, farming and pets.

The short promotion video below highlights the experience, with special guest stars at 0:16 and 0:25 (my kids).

Outside of the venue, Winchester Science Centre actively enagage the community, including local schools, science festivals, and freely available online STEM activities, supported by their passionate team of STEM Ambassadors.

If you’re interested in STEM and live in the south of England, I highly recommend a trip to Winchester Science Centre (SO12 1HZ).

Alternatively, keep an eye on their social channels (Instagram, Twitter, Facebook) for all things STEM!

Security Policy Templates

Mon, 18 Jul 2022 00:00:00 -0500

In my previous article, I shared a range of cybersecurity resources.

In addition, I wanted to share a series of security policy templates developed and posted by SANS.

https://www.sans.org/information-security-policy/

SANS, launched in 1989, is a cooperative for information security thought leadership. Their mission is to empower cybersecurity professionals with practical skills and knowledge.

The policy templates are a great starting point for any organisation looking to establish or mature a cybersecurity programme. The key categories include:

SANS also promote community contributions via a form, although it would be nice if this was handled via a version control system.

Cybersecurity Resources

Sat, 16 Jul 2022 00:00:00 -0500

In May, my role was expanded to include Chief Information Security Officer (CISO). Over the past few months, I have been familiarising myself with all things cybersecurity.

Therefore, I thought I would share my resource list. All links are safe and legal (ethical). The majority are free or include free trials.

Cybersecurity Web Resources/Newsletters

Cybersecurity YouTube/Podcasts

Cybersecurity (Ethical Hacking) Training

Cybersecurity Search Engines

Have I Been Pwned (HIBP) - Account Compromise
CVE Details - Common Vulnerabilities and Exposures Database
ExploitDB - Exploit Database
Dehashed - Leaked Credentials

Cybersecurity Web App Software Tools

Nmap - Port Scanning
OWASP ZAP - Web App Scanner
Burp Suite - Web Security Testing
Dirsearch - HTTP Bruteforcing
Amass - Network Mapping
TruffleHog - Secret Scanning
sqlmap - SQL Injection Flaws

Cybersecurity Hardware

I also recommend you follow Daniel Kelley on LinkedIn. He posts a wide range of resources related to Cybersecurity. A few items from my list were inspired by his posts.

Corporate Separation - Update

Thu, 14 Jul 2022 00:00:00 -0500

In January, World Wide Technology (WWT) published a case study highlighting our journey to complete a corporate separation in just two years, described by many as an impossible mission.

Earlier this week they updated the article with a video interview.

Corporate separations do not happen overnight; they take an incredible number of resources, time, strategic planning, and rapid delivery.

Thanks to WWT for the partnership!

NIST Cybersecurity Framework

Thu, 07 Jul 2022 00:00:00 -0500

In May, my role was expanded to include Chief Information Security Officer (CISO). Since that date, I have been building trust with the team, as well as connecting with internal and external stakeholders. I have also been re-educating myself, going back to basics with key cybersecurity approaches, methodologies and frameworks.

This article provides a summary of the National Institute of Standards and Technology (NIST) Cybersecurity Framework (CSF), also known as the NIST CSF.

The NIST CSF is a set of guidelines for mitigating organisational cybersecurity risks, based on established standards, guidelines, and practices. It includes three primary components:

Core: Desired cybersecurity outcomes organised in a hierarchy and aligned to more detailed guidance and controls.
Profiles: Alignment of an organisation’s requirements and objectives, risk appetite and resources using the desired outcomes of the NIST CSF Core.
Implementation Tiers: A qualitative measure of organisational cybersecurity risk management practices.

The NIST CSF is intended to be a “living document” that is continuously refined, improved, and evolved, hopefully keeping pace with technology and threat trends, whilst integrating lessons learned.

The full NIST CSF v1.1 can be downloaded below:

NIST CSF v1.1

The Framework Core includes 5 functions (listed below), covering 23 categories and 108 subcategories, with 6 informative references. The informative references include ISO 27001, COBIT, NIST SP 800-53, ANSI/ISA-62443, and the Council on CyberSecurity Critical Security Controls.

Identify: Develop the organizational understanding to manage cybersecurity risk to systems, assets, data, and capabilities.
Protect: Develop and implement the appropriate safeguards to ensure the delivery of critical infrastructure services.
Detect: Develop and implement the appropriate activities to identify the occurrence of a cybersecurity event.
Respond: Develop and implement the appropriate activities to take action regarding a detected cybersecurity incident.
Recover: Develop and implement the appropriate activities to ensure resilience and restore capabilities impacted by a cybersecurity incident.

The Framework Core is available as a downloadable CSV, linked below:

NIST CSF Core v1.1

The Framework Profile is a list of outcomes that an organisation has chosen from the categories and subcategories, based on its needs and risk assessments.

Typically, an organisation would start by developing the “Current Profile” which describes the cybersecurity activities and what outcomes are being achieved.

This leads to the development of a “Target Profile”, resulting in a gap analysis and defined actions to mature from the “Current Profile” to the “Target Profile”.

Finally, the Implementation Tiers, provide a consistent structure to benchmark cybersecurity risk management practices, helping organisations develop plans to improve their cybersecurity posture.

It is essential to note the Implementation Tiers are not designed to be a maturity model, instead providing visibility across cybersecurity risk management and operational risk management processes.

The four Implementation Tiers are outlined below:

Tier 1 (Partial): Organisations with limited security processes defined and implemented, resulting in low cybersecurity maturity, lacking sponsorship and prioritisation.
Tier 2 (Risk Informed): Organisations that understand the risks and are actively addressing compliance requirements. However, they are not addressing broader (organisation-wide) security concerns and/or policy gaps.
Tier 3 (Repeatable): Organisations that have executive-approved risk management and cybersecurity best practices defined and implemented. Organisations that are prepared to respond to cybersecurity threats, and risks, whilst proactively addressing vulnerabilities.
Tier 4 (Adaptive): Organisations that utilise advanced adaptive cybersecurity practices, where behaviours and/or events are proactively analysed to protect from or adapt to threats before they happen. In parallel, organisations that continuously assess risk and automatically provide proportional enforcement as required.

For more details regarding the NIST CSF, I recommend reviewing the “NIST Cybersecurity Framework Explained” virtual session from Tom Conkle and Kelly Hood. It was presented at the RSA Conference in 2018 but is still relevant today.

The NIST CSF is a powerful resource for SMB and Enterprise organisations across the globe, providing a robust framework to identify cybersecurity outcomes and a methodology to assess and manage those outcomes.

Journey to Hybrid Multi-Cloud

Tue, 21 Jun 2022 00:00:00 -0500

Yesterday, Calum Bell and I presented at HashiConf Europe 2022, which is now available online for on-demand viewing.

The presentation focused on our journey to Hybrid Multi-Cloud, specifically, Microsoft Azure and Google Cloud Platform (GCP), automated using HashiCorp Terraform Cloud.

I provided context regarding our business, IT ecosystem and hosting architecture.

Business Context
IT Ecosystem (01:05)
Hybrid Multi-Cloud Architecture (01:38)

Calum covered our automation journey, including the analysis, design and implementation.

Automation Context (02:38)
Analysis and Design Phase (03:22)
Implementation Phase (09:19)
The Future (14:20)

All presentations, including the keynotes, are available online for on-demand viewing. However, access does require a free HashiCorp account.

Thanks to the team at HashiCorp for hosting a great event! Calum and I welcome any questions from the community.

Steam Deck

Fri, 10 Jun 2022 00:00:00 -0500

On the 16-JUL-2021 at 18:00 BST, the highly anticipated Steam Deck from Value become available for pre-order.

Knowing there would be limited availability, compounded by the silicon shortage and supply chain constraints, I was primed to place my pre-order, ready for the scheduled December 2021 release date.

Unsurprisingly, the Steam Store struggled to handle the increased traffic, with many users reporting issues. Thankfully, my order was completed successfully, with an order time of 18:01 BST, approximately one minute after the pre-order window opened.

Eleven months later, I finally received my Steam Deck!

Yes, it has taken Value nearly one year to meet the demand from the first minute of pre-orders.

In fairness, Value is not to blame for the long lead time. Compared to other “high profile” hardware launches (e.g. PS5, Xbox Series S/X), I believe Valve has done a good job, with a verified pre-order process, clear expectations and consistent weekly deliveries.

My first impression upon opening the box, the Steam Deck is massive but surprisingly lightweight. The image below shows the Steam Deck alongside my other handheld consoles, specifically the Nintendo Switch (HAC-001), PlayStation Vita 1000 (OLED), and Anbernic RG351MP.

The Steam Deck makes the other consoles look tiny by comparison, even the Nintendo Switch, was considered large at launch.

It is worth briefly highlighting the Anbernic RG351MP, which is a surprisingly good retro console, first released in 2020. It is incredibly compact, with a robust (metal) construction and the highly versatile RK3326 Quad-Core processor, capable of running emulators from Game Boy through to PlayStation (even Nintendo 64 and Dreamcast, if you are willing to accept some slowdown).

The build quality of the Steam Deck is very good, especially when you consider the aggressive £349 starting price. I ordered the 256GB version, which comes with a faster NVMe SSD, priced at £459.00.

In my opinion, the Steam Deck does not feel as refined as the original PlayStation Vita 1000 (OLED), which (in my opinion) was years ahead of its time. However, when you consider the impressive specification (AMD Zen 2 4C/8T, AMD RDNA 2 GPU and 16GB LPDDR5 Memory) and the massive array of inputs (list below), it is hard not to be impressed.

2x Analog Sticks with Capacitive Touch
2x Trackpads
2x Shoulder Buttons (L1, R1)
2x Analog Triggers (L2, R2)
4x Grip Buttons (L4, R4, L5, R5)
D-Pad
A/B/X/Y Buttons
6-axis IMU (3-axis Accelerometer, 3-axis Gyroscope)
View/Menu Buttons
Steam Button
Quick Access Button
Volume +/− Buttons
Power Button

All the buttons are reassuringly tactile, with only the trackpads feeling a little “cheap”, although only when compared to a modern multi-touch touchpad.

The Analog Sticks with Capacitive Touch are a particular highlight, easily comparable to the current generation of consoles (e.g. PlayStation DualSense Wireless Controller). The fact that they are also modular and therefore user-serviceable is also a major positive for “right to repair”.

The Steam Deck runs a modified version of Arch Linux called SteamOS v3.0, which natively integrates the Proton compatibility layer.

I have previously written about Proton, which is a fork of the popular Wine compatibility layer.

The name Wine stands for “Wine Is Not an Emulator”, which is an important distinction, as it aims to translate (not emulate) Windows API calls into native POSIX calls. This approach eliminates the performance and memory penalties of methods such as emulation and virtualisation.

In short, Proton is the “magic” behind the Steam Deck, which allows games designed for Windows to run on Linux at native performance.

The SteamOS default user interface is optimised for the Steam Deck, providing quick access to the Steam Store, personal games library and settings.

However, you can also switch to a full Linux desktop, which enables the installation of third-party software via Flatpak applications.

This unprecedented flexibility unlocks the true potential of the Steam Deck, allowing for the installation of games from third-party stores, such as the (Epic Games Store, GOG, etc.)

It is even possible to install non-gaming software (e.g. Browsers, Productivity, Collaboration), assuming it is available as a Flatpak.

For example, I installed the Heroic Games Launcher, which provides native Linux access to the Epic Games Store and GOG, including your associated library of games.

The Heroic Games Launcher can be downloaded, installed and maintained from the integrated “Discover Software Center” and then added to your Steam library as a “Non-Steam Game”.

If you want an extra level of integration, the “HeroicBashLauncher” creates launch scripts for installed Epic and GOG games, allowing them to be added as individual games within Steam (meaning you can also configure per-game settings).

With Steam and the Heroic Games Launcher, alongside Microsoft xCloud and EmuDeck, the scope of games capable of running on the Steam Deck is massive!

As an example, over the years, I have taken advantage of many gaming offers (many free), resulting in a substantial games collection. With these games locked to the desktop, I rarely get time to play. The Steam Deck has the potential to change that, providing greater flexibility when travelling, etc.

Expect additional updates regarding the Steam Deck. Knowing that it is an x86-64 Linux-based PC “under the hood”, I am excited to see how far it can be pushed.

Framework Performance

Wed, 01 Jun 2022 00:00:00 -0500

Over the past month, I have been testing the Framework Laptop, from Framework Computer (which for brevity, I will simply reference as Framework).

I shared my initial impressions and setup process in the following two articles.

As a brief reminder, Framework set out to deliver a laptop that encourages home servicing and incremental upgrades.

As highlighted in my last article, I selected Fedora 36 as my operating system.

The Framework Laptop is fully Windows 10/11 compatible, however, I am a Linux user and the “open” philosophy of Framework felt like a perfect match.

In this article I plan to share my experience using the laptop over the past month, covering performance, reliability and usability.

It should be noted that as I am running Linux, my impressions may not reflect the Windows experience, which has different performance and power consumption characteristics.

With that said, I am pleased to report that Framework has published handy guides that cover the installation process for popular Linux distributions.

With Fedora 36, everything works out of the box, including common “pain-points” such as the audio, webcam and fingerprint reader.

It is, however, important to ensure you are running the latest BIOS, which at the time of writing is version 3.07.

At this time, the BIOS update cannot be triggered from within Linux. However, Framework is actively working to implement LVFS (Linux Vendor Firmware Service).

As a result, the BIOS must be updated manually, by downloading the relevant files from the Framework Help Center.

Once downloaded, updating is a simple three-step process:

Extract contents of “.zip” to a FAT32 formatted USB drive.
Disable Secure Boot in BIOS (F2 > Security > Secure Boot).
Boot the system while tapping F12, and select the USB drive.

The remainder of this article assumes BIOS version 3.07.

Performance

The specification of my Framework Laptop can be found below.

Framework Laptop DIY Edition
Intel i7-1185G7 4.80GHz (4C/8T)
64GB Crucial DDR4 PC4-25600C22 3200MHz RAM
1TB Western Digital Black SN850 NVMe (7GB/s Read)
Intel Iris Xe Graphics
13.5-inch LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

Knowing that the Framework Laptop is a standard x86-64 architecture, running an Intel 11th Generation processor, the performance is highly predictable.

For example, any generic benchmark and/or review of the Intel i7-1185G7 will provide a good insight into the performance.

The main differentiator would be any limitation regarding thermal throttling, which I am pleased to report is not something I have experienced with the Framework Laptop.

To provide some numbers, the Geekbench 5 score was 1727/6134.

I also tested Docker, leveraging the build process outlined in my article “Docker and Apple Silicon”.

The build was completed in 54 seconds, which interestingly outperformed the Dell XPS 17 and Ryzen 3950X custom-built PC (although this was running Windows and WSL2).

Overall, the performance of the Framework Laptop is very good. Not groundbreaking, but consistent with other laptops with an equivalent specification.

Finally, it is worth mentioning that Framework recently announced a series of Intel 12th Generation mainboards, which can be installed as an upgrade.

Intel i5-1240P (12C/16T)
Intel i7-1260P (12C/16T)
Intel i7-1280P (14C/20T)

With the combined benefits of the newer 12th Generation architecture and higher core count, I would expect to see a decent performance improvement over the current 11th Generation options.

Reliability

So far, I have been impressed with the reliability of the Framework Laptop running Fedora 36.

I have not experienced any errors or unforeseen issues caused by hardware or software. Although this should be a baseline expectation, there has been a worrying trend of poor quality control with consumer electronics, even from “big brands” such as Dell.

Knowing that Framework is a new company, combined with the additional complexity of designing a modular laptop that can be serviced at home, I would have accepted a few “issues”. Therefore, it is great to see this level of quality in their first consumer product.

The only area I would highlight relates to the display hinge, which (in my opinion) is not rigid enough, which can result in a minor wobble when typing.

Interestingly, it would appear this was a common view and Framework recently released a new hinge kit, which includes a more rigid (4.0kg up from 3.3kg) mechanism.

This is a great example of listening to your customers and the unique advantage of a modular design, where individual parts can be replaced.

Finally, it is important to highlight the battery life, which I would describe as “ok”.

In theory, the laptop can achieve 10 hours of usage. However, I do not believe I have ever achieved this outcome. In my experience, I would guess between 6 and 8 hours.

However, when using Linux, it is “sleep” mode that causes the biggest concern. By default, using Fedora 36, the Framework Laptop battery can drain as much as 15% per hour while in sleep.

This is far from ideal, likely caused by the fact that Fedora 36 (at this time) cannot take advantage of the full sleep state of the Intel processor.

To mitigate this issue, I switched the default sleep state from “s2idle” to “deep”, using the commands highlighted in my previous article. This change reduces the battery drain to approximately 3% per hour, which is not perfect, but manageable.

As a result, I do not believe I could rely on the Framework Laptop as an “all-day battery” laptop, which is a shame, as I have been spoilt by the latest Apple MacBook Pro running Apple Silicon.

Usability

Overall, the Framework Laptop has been a pleasure to use, either used as a standalone laptop or connected to my Samsung C49RG90 49-inch Super Ultra-Wide monitor.

The keyboard, modular expansion cards, 3:2 aspect ratio, and display flexibility are particular highlights.

The touchpad, speakers, webcam and microphone are all good, certainly not the best I have used, but perfectly viable for daily use. I do like the physical buttons to disconnect the webcam and microphone, providing additional reassurance for privacy-conscious users.

The display itself and fingerprint reader are my only frustrations. The display aspect ratio (3:2) and resolution (2,256 x 1,504) are great. However, it is highly reflective.

The fingerprint is ok, but (in my experience) unreliable when compared to Apple TouchID. This might be due to the Linux driver vs. the hardware, which I have not been able to test with Windows.

Finally, it is worth reiterating the primary selling point of the Framework Laptop. In the event of a hardware issue, I can simply swap the individual component.

Conclusion

In conclusion, I can’t help but love the Framework Laptop.

Is it perfect? No.
Is it the best laptop I own? No (Apple MacBook Pro M1 Max).
Do I still love it? Yes.

The vision and ambition of Framework as a company is infectious and they have done an amazing job delivering a product that redefines how we think about a laptop, with a focus on the user and sustainability.

I am eager to support their cause and excited to see what the future holds!

Framework and Fedora

Fri, 27 May 2022 00:00:00 -0500

Earlier this month, I shared my first impressions of the Framework Laptop.

As a reminder, Framework set out to deliver a laptop that encourages home servicing and incremental upgrades. I ordered the “DIY Edition”, with the following specification:

Framework Laptop DIY Edition
Intel i7-1185G7 4.80GHz (4C/8T)
64GB Crucial DDR4 PC4-25600C22 3200MHz RAM
1TB Western Digital Black SN850 NVMe (7GB/s Read)
Intel Iris Xe Graphics
13.5-inch LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

The assembly process was incredibly straightforward, supported by smart design decisions from Framework, specifically, well-organised internals, labels and instructions for each serviceable part, as well as the inclusion of a handy T5/PH0 double-sided bit screwdriver.

The build quality of the laptop is very good, but not what I would consider premium. The Framework Laptop does not have the same dense, refined feel as an Apple MacBook or Dell XPS. Instead, I would compare it to a mid-range Dell, with a focus on practical usage vs. aesthetics.

For example, the chassis is made from 50% post-consumer recycled (PCR) aluminium, but it is not overly rigid, resulting in some screen flex and wobble when typing. The same can be said for the trackpad, which lacks the “precision feel” of a Dell XPS or Razer Blade.

With that said, these are minor complaints that do not impede the daily usage of the laptop.

Since my purchase and following their promise, Framework has released newly upgraded parts and open-source accessories.

The most significant is the Intel 12th Generation mainboard, which remains compatible with the original chassis.

There have also been some interesting community modifications, using the published schematics to 3D print accessories, such as a standalone case for the mainboard.

In short, the hardware continues to impress and it is great to see Framework deliver new updates/parts, helping to build trust in their promise to enable home servicing and incremental upgrades.

Framework and Fedora

Complimenting the “open” nature of the hardware, I felt it was only right to install an open-source operating system.

The Framework Laptop is fully Windows 10/11 compatible and the likely option for most users. However, as a Linux fanatic, there was only one choice for me!

Initially, I planned to install Pop!_OS 22.04, which includes Linux Kernel 5.16.19, delivering native compatibility for the Framework Laptop.

Frustratingly, I encountered an issue with the WiFi driver. In theory, this can be fixed by installing the backport Intel WiFi driver package using the command below.

sudo apt install backport-iwlwifi-dkms

However, the included Intel AX210 vPro WiFi simply refused to connect. I tried multiple workaround options (following guides on Reddit), but no luck.

Although I have grown to love Pop!_OS, I decided to switch to my old faithful, Fedora. The timing was good, as Fedora 36 had just been released and I was eager to check it out.

I am pleased to report the installation of Fedora 36 was perfect, with full drive encryption and every driver, including WiFi and the fingerprint reader, detected natively.

This is a great example of how good Linux can be, with no need to install/configure third-party proprietary drivers, etc.

I suspect my issue with Pop!_OS was simply a bug, which I am confident will be resolved soon.

Once up and running, I installed my default software. Thanks to the growing popularity of immutable operating systems such as Silverblue and the recent launch of the Steam Deck, the majority of my software is available as a flatpak.

Therefore, the installation was as simple as running the following commands.

flatpak remote-add --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo

flatpak install flathub org.onlyoffice.desktopeditors
flatpak install flathub com.simplenote.Simplenote
flatpak install flathub com.visualstudio.code
flatpak install flathub org.gimp.GIMP
flatpak install flathub org.kde.krita
flatpak install flathub org.flozz.yoga-image-optimizer
flatpak install flathub com.obsproject.Studio
flatpak install flathub com.getpostman.Postman
flatpak install flathub org.videolan.VLC
flatpak install flathub fr.handbrake.ghb
flatpak install flathub org.cryptomator.Cryptomator
flatpak install flathub org.godotengine.Godot
flatpak install flathub com.prusa3d.PrusaSlicer
flatpak install flathub com.valvesoftware.Steam
flatpak install flathub com.heroicgameslauncher.hgl
flatpak install flathub com.discordapp.Discord
flatpak install flathub us.zoom.Zoom
flatpak install flathub com.microsoft.Teams

Outside of flatpak, I installed Docker Engine and Docker Desktop. Fedora includes Podman, which is an excellent container engine.

However, I am most familiar with Docker and haven’t invested the time (yet) to migrate my workloads.

Getting Docker up and running is a little more involved, achieved via the following commands.

sudo dnf -y install dnf-plugins-core

sudo dnf config-manager \
    --add-repo \
    https://download.docker.com/linux/fedora/docker-ce.repo

sudo dnf install docker-ce docker-ce-cli containerd.io docker-compose-plugin

sudo systemctl start docker

Docker Desktop is available as an “rpm” download, which I installed using the following command.

sudo dnf install ./docker-desktop-<version>-<arch>.rpm

The login for Docker Desktop relies on the “pass” password manager to store credentials as gpg2-encrypted files.

Therefore, before authenticating via Docker Desktop, “pass” must be initialised.

gpg --generate-key

pass init <key>

1Password, is my prefered password manager. It is available as an rpm package, installed via the following commands.

sudo rpm --import https://downloads.1password.com/linux/keys/1password.asc

sudo sh -c 'echo -e "[1password]\nname=1Password Stable Channel\nbaseurl=https://downloads.1password.com/linux/rpm/stable/\$basearch\nenabled=1\ngpgcheck=1\nrepo_gpgcheck=1\ngpgkey=\"https://downloads.1password.com/linux/keys/1password.asc\"" > /etc/yum.repos.d/1password.repo'

sudo dnf install 1password

Finally, I install a few “tweaks” to ensure reliable media playback and font compatibility.

Starting with RPM Fusion, which is a repository of add-on packages for Fedora. It includes some useful packages that are not directly included with Fedora.

The following command enables the “free” and “non-free” repositories.

sudo dnf install https://mirrors.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm https://mirrors.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm

With RPM Fusion enabled, the following commands install common multimedia libraries.

sudo dnf install gstreamer1-plugins-{bad-\*,good-\*,base} gstreamer1-plugin-openh264 gstreamer1-libav --exclude=gstreamer1-plugins-bad-free-devel

sudo dnf install lame\* --exclude=lame-devel

sudo dnf group upgrade --with-optional Multimedia

Next is fonts, where I install the standard Microsoft fonts for improved Microsoft Office compatibility.

sudo dnf install mscore-fonts-all -y

sudo dnf install curl cabextract xorg-x11-font-utils fontconfig

sudo rpm -i https://downloads.sourceforge.net/project/mscorefonts2/rpms/msttcore-fonts-installer-2.6-1.noarch.rpm

Throughout the setup process, I noticed the sleep/suspend state was not ideal, essentially resulting in a high battery drain whilst the laptop lid was closed.

By default, Fedora does not support hibernate (although it can be configured). Therefore, the default sleep state is “s2idle”, which is not well optimised. This can be confirmed using the following command.

cat /sys/power/mem_sleep

Although not perfect, I switched the sleep state to “deep” by modifying the file ‘/etc/default/grub’, inserting the parameter ‘mem_sleep_default=deep’ at the end of the line containing the ‘GRUB_CMDLINE_LINUX’ string.

The final string should read:

GRUB_CMDLINE_LINUX="mem_sleep_default=deep"

Once added, I rebuilt the grub file using the following command (specific to UEFI).

sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg

Upon reboot, the sleep state should now default to “deep”, which in my testing significantly reduces the battery drain with the lid closed, with the only caveat being a slightly slower resume time (approximately 15 seconds).

That’s it! My Framework Laptop is now fully set up, running Fedora 36.

Overall, the process was fairly painless, primarily thanks to the “Linux friendly” hardware choices and the modern Linux Kernel (Linux Kernel 5.17) included as part of Fedora 36.

Over the coming weeks, I plan to put the Framework Laptop through its paces. Therefore, be sure to stay tuned for updates regarding performance, reliability and usability.

HashiConf Europe 2022

Fri, 20 May 2022 00:00:00 -0500

Update - The HashiConf Europe 2022 presentation is now available to view online.

Next month, Calum Bell and I will be presenting at HashiConf Europe 2022.

HashiConf Europe 2022, hosted by HashiCorp, includes over forty keynote and technical sessions, as well as product demos, virtual labs and social events.

Due to COVID-19, HashiConf Europe 2021 was virtual, but in 2022 the conference will embrace hybrid working, with physical (Amsterdam) and virtual attendance.

The video below provides a brief recap of HashiConf Europe 2021, highlighting the type of content presented.

Our session will focus on our journey to Hybrid Multi-Cloud, specifically, Microsoft Azure and Google Cloud Platform (GCP) using HashiCorp Terraform Cloud.

We plan to provide some insight into our business and our rationale for embracing Hybrid Multi-Cloud.

We will then outline our developer experience, which prioritises product team autonomy and proactive controls that help ensure production releases are scalable, secure and meet our expectations regarding quality, privacy and sustainability.

The session will be made available for on-demand viewing; therefore, I stay tuned for a follow-up article.

If you would like to engage live, the conference takes place between the 20th and 22nd of June and is free to all virtual attendees.

Framework Laptop

Thu, 05 May 2022 00:00:00 -0500

In January 2020, Framework Computer (which for brevity, I will simply reference as Framework) announced its intent to produce a laptop that respected the consumer’s right to repair and upgrade.

Conventional wisdom, adopted by most computer manufacturers (Apple, Dell, HP), accepts that producing repairable products will make them thicker, heavier, uglier, less robust, and more expensive.

Framework believes they can reset this perspective, delivering a modern, performant, desirable, and cost-effective laptop, with a modular design that encourages home servicing and incremental upgrades.

Like many across the technology industry, I was excited by this prospect, which is pro-consumer and environmentally friendly.

For example, the world produces more than fifty million tons of e-waste per year. On average, that is 6kg per person, per year. This horrific statistic is a direct result of the monolithic, anti-environmental stance taken by companies such as Apple, etc.

Don’t get me wrong, I use (and love) Apple products, but I also recognise the need to change, which will only happen if consumers take a stand and support companies like Framework.

Proposed government legislation, such as the right to repair, would also help, forcing manufacturers to provide the practical means for consumers to repair their devices.

Framework Laptop

In July 2021, Framework shipped its first product, the Framework Laptop, to the US and Canada. This was followed by the UK launch in December 2021.

Last month, I received my Framework Laptop, which included the following specification:

Framework Laptop DIY Edition
Intel i7-1185G7 4.80GHz (4C/8T)
64GB Crucial DDR4 PC4-25600C22 3200MHz RAM
1TB Western Digital Black SN850 NVMe (7GB/s Read)
Intel Iris Xe Graphics
13.5-inch LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

I selected the “DIY Edition”, which allows you to customise your order, shipped as components for home assembly.

My selected specification is the highest performance available from Framework, costing a total of £2,093.00 (including the £49 USB-C power adapter).

The price to performance ratio is acceptable, certainly not the cheapest for the specification, but competitive with other premium laptop manufacturers (e.g. Dell, HP, Lenovo, etc.)

However, the “magic” of the Framework Laptop is not the price or performance, it is the promise of a highly modular laptop that can be serviced/upgraded incrementally over time.

Building the Framework Laptop

Outlined below are my initial thoughts, covering the unboxing and building of the Framework Laptop. I plan to publish a follow-up article, covering the usage, etc.

The Framework Laptop arrived in a single cardboard box, which included three smaller boxes (laptop, custom components and power adapter).

Inside the main box was the laptop itself. Even with the “DIY Edition”, the mainboard, display, keyboard, touchpad and battery were already installed.

However, these components are sold separately via the Framework Marketplace.

The Framework Screwdriver is a combination spudger and screwdriver that comes with a magnetically attached T5/PH0 double-sided bit.

The second box contained my custom configured components, specifically the Memory, Storage and Expansion Cards. It would appear this is a standard box, with empty compartments for additional components.

The Framework Laptop itself includes a 13.5-inch display, with a 3:2 aspect ratio. The exterior build quality is good, comparable to a mid-range PC laptop, although not quite achieving the same level of refinement as an Apple MacBook or a Dell XPS.

Although this reads like a criticism, the build quality is perfectly acceptable, acknowledging that Framework was founded in 2020, compared to the long heritage of manufacturers such as Apple (1976), Dell (1984), and HP (1939).

The keyboard and touchpad are both great! The keyboard delivers excellent travel and responsiveness. The touchpad is a buttonless design (not haptic/static) providing a large surface area similar to a Dell XPS 13.

The only brand awareness on the inside of the laptop is the Framework logo/text on the display hinge, which is tastefully debossed.

A wider view of the keyboard and touchpad highlights the standard QWERTY layout in UK format.

The official Framework power adapter is rated at 60W, with a detachable power and USB-C cable. The power brick is very compact thanks to the use of gallium nitride (GaN).

A closer look at the Framework Screwdriver, which is the only tool you need to service/upgrade your Framework Laptop. The inclusion of this tool is important, as it encourages the consumer to take action.

The Framework Laptop is easy to open, simply loosen the five T5 screws on the bottom. At this point, the keyboard base can be lifted, revealing the laptop internals. All screws are connected to the base (spring-loaded), which is great, as it means they cannot be lost.

Each serviceable component within the Framework Laptop is easily accessible and clearly signposted, with a handy QR code providing direct access to the specific service/replacement guide (e.g. Battery, Memory, Storage, etc.)

The guides are also excellent, similar to what I have seen from Prusa, providing concise step-by-step information, including photos, videos and user feedback.

The attention to detail on the inside of the Framework Laptop is impressive, arguably more impressive than the outside. The team should be commended for this innovation.

The memory installation process was exceptionally simple, exactly as it should be.

The same can be said for the storage, with the same T5 screw-type being used to mount the SSD.

The touchpad is also serviceable, although likely a little more challenging to remove. The guide highlights the difficult as “Moderate”. Each serviceable component includes a difficulty level, which helps set appropriate expectations.

The final part of the build was to insert the expansion cards. The Framework Laptop includes four expansion ports, which all connect via USB-C.

There are currently six expansion cards available, with the promise of more in the future.

I selected, 2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), and 1x HDMI 2.0b.

Once inserted, the expansion card can be removed by pressing the button on the bottom of the laptop.

Overall, I commend the Framework team for leading by example and showing the world that serviceable/upgradable laptops are still possible in 2022!

Steve Jobs once said:

“Making an insanely great product has a lot to do with the process of making the product, how you learn things and adopt new ideas and throw out old ideas.”

I believe this statement epitomises the Framework Laptop, which is a tangible example of innovation that challenges the status quo.

I am excited to see how the Framework Laptop performs over time, attacking my common workloads. Expect a follow-up article soon.

CTO vs. CISO

Sun, 01 May 2022 00:00:00 -0500

Since 2018, I have held the role of Chief Technology Officer (CTO).

Last month, my role was expanded, combining the roles of Chief Technology Officer (CTO) and Chief Information Security Officer (CISO).

The expansion was the result of an unforeseen personal change and not necessarily part of my career plan.

However, sometimes circumstances present an opportunity and it was certainly a gratifying “vote of confidence” in my ability. Therefore, I willingly accepted, excited by the prospect of personal development and the chance to broaden my value contribution.

With that said, I did have some reservations regarding the impact of combining the CTO and CISO, both personally and from a business perspective.

Therefore, I thought I would share my thinking, highlighting some of the benefits and concerns.

As a starting point, I feel it is important to baseline the responsibilities of the CTO and CISO.

It should be noted that these responsibilities are not consistently defined. For example, depending on the company, the CTO role can cover everything from technology operations management to technology innovation (and sometimes both).

Outlined below are the responsibilities of the CTO and CISO at my company.

Chief Technology Officer (CTO)

The Chief Technology Officer (CTO) is the executive responsible for the overarching technology strategy, including investments, architecture, engineering and external engagement.

Business Development: Partner with the business to build a strategy that supports the company objectives. Help to identify and realise opportunities, including new digital/data business models.
IT Vision and Strategy: Position IT as a differentiator, delivering a secure, reliable, efficient IT Ecosystem and new value through the creation of innovative digital/data products, services and insights.
Technology Governance: Establish, embed and maintain a framework of authority and accountability that defines and controls the outputs, outcomes and benefits of Information Technology, ensuring prioritised initiatives have the required structure, sponsorship and funding to succeed.
Technology Investments: Establish and maintain the technology investment portfolio, including the executive relationships with strategic partners, aiming to maximise the return on investment.
Architecture: Lead the global architecture community, establishing, embedding and maintaining a framework for the design, development, and implementation of IT/OT solutions. Includes the leadership of Business Process, Solution, Data, and Domain architects, responsible for the creation and maintenance of architecture principles, positioning, methods and models, including technical enforcement mechanisms.
Engineering: Lead the global engineering community, enabling the design, development, and implementation of IT/OT solutions, alongside rapid development of discovery (experimentation) initiatives. Includes the leadership of Solution, Data, Site Reliability, and DevOps engineers.
Innovation and External Engagement: Identify and promote technology opportunities that position the company as a market leader in Science, Technology, Engineering and Math (STEM), supporting talent acquisition and external collaboration.
Coaching and Mentoring: Engage in the community (internal and external) across multiple channels, looking to share, educate and inspire. Support and promote the recruitment and personal development of individuals following a career path in STEM.

Chief Information Security Officer (CISO)

The Chief Information Security Officer (CISO) is the executive responsible for developing and implementing an information security program, covering cyber, data and information security.

Cybersecurity Governance: Establish, embed and maintain a framework of authority and accountability for cybersecurity across Information Technology (IT), Operational Technology (OT) and relevant Business Functions, ensuring prioritised initiatives have the required structure, sponsorship and funding to succeed.
Cybersecurity Risk: Partner to analyse, measure and understand cybersecurity risk, identifying potential threats and vulnerabilities following a defined risk framework, including strategies to mitigate the likelihood.
Cybersecurity Compliance: Identify and maintain compliance with relevant global cybersecurity laws, regulations, and industry standards. Establish and maintain roles and responsibilities with key Business Functions, specifically Finance, Legal and Privacy.
Cybersecurity Architecture: Ensure cybersecurity policies and standards are understood and enforced throughout the design, development, and implementation of Information Technology (IT) and Operational Technology (OT) solutions. Includes the leadership of Information Security architects, responsible for the creation and maintenance of architecture principles, positioning, methods and models, including technical enforcement mechanisms.
Cybersecurity Operations: Continuous analysis of cybersecurity threats, incidents and investigations, maintaining the global incident response plan that defines the detect, respond and recovery processes.
Cybersecurity Awareness and Education: Deliver and measure cybersecurity awareness and education to improve the acumen of Elanco stakeholders, establishing a culture of security.
Cybersecurity Technology Foundations: The procurement, design, implementation and support of specific cybersecurity technologies/services, facilitating cybersecurity policies, standards, controls and procedures.
Executive Engagement: Represent cybersecurity internally and externally, partnering closely with the Executive Committee, Audit Committee and Board of Directors.

Combined CTO / CISO

As highlighted by the responsibilities, there are natural points of convergence, specifically architecture and engineering when targeting Security by Design.

However, each role also includes a range of unique responsibilities, which have the potential to be time-consuming, commonly resulting in two dedicated roles.

Outlined below are the advantages and concerns associated with a combined CTO / CISO role.

Advantages:

Zero Trust: As highlighted in my article “Zero Trust”, the advent of digital/data business models has resulted in new threat vectors. Zero Trust is a fundamentally different approach to IT security, moving away from the traditional “moat/castle” strategy. It places a strong emphasis on Identity, the Principle of Least Privilege, and Securing at Source. I believe a combined CTO / CISO role could help accelerate the adoption of Zero Trust by positioning a unified strategy, ensuring discipline and removing bureaucracy.
IT Ecosystem Knowledge: The CTO should have an unparalleled understanding of the IT Ecosystem. Although certain roles have greater domain-specific expertise, very few roles have the full end-to-end understanding, with the ability to deep dive where required. This insight provides a unique appreciation of the strengths and weaknesses, which could help identify and prioritise risks.
Technical Expertise: Although technical depth is not a requirement of the CISO role, the technical expertise commonly possessed by the CTO could prove invaluable. For example, “code-level” knowledge when dealing with a software-defined IT ecosystem could help identify opportunities and troubleshoot incidents, whilst ensuring appropriate context when assessing risk.
Delivery/Ops: Security by Design requires product/project teams to embed security into their daily activities. This can be a challenge for dedicated Information Security teams, where roles and responsibilities can become confused or ignored. The established relationships and credibility between the CTO and Delivery/Ops teams should help bridge this gap, promoting and embedding the mission of Information Security.

Concerns:

Equilibrium: The CTO and CISO roles work best when balancing each other. The CTO should be progressive, pushing the boundaries of what is possible. The CISO should constructively challenge, helping to manage any associated risk. With the two roles combined, a new approach would need to be defined to self-regulate.
Governance, Risk and Compliance (GRC): A critical part of the CISO role is Governance, Risk and Compliance (GRC), establishing a framework of authority and accountability that defines and controls the outputs, outcomes and benefits of Information Security. This is a process-led (not technology-led) responsibility, requiring a different skill set to be successful.
Information Security Team: Succession in Information Security is commonly associated with the CISO role. By combining the CTO / CISO role, the expectations regarding expertise and experience are reset. This would likely reduce the viable candidates, potentially impacting morale and creating retention challenges.
Capacity: As highlighted, the CTO and CISO roles require a significant time investment to be successful as they both include a wide range of highly visible responsibilities. A combined CTO / CISO role will undoubtedly apply additional individual pressure, potentially impacting work/life balance and wellbeing. Certainly, something to consider when looking to ensure a healthy “long-term” career.

Conclusion

Looking across the industry, it is common for start-ups and small/medium businesses to have a combined CTO / CISO role. However, it becomes less common at the enterprise level.

In a perfect world, I do believe these two roles are best delivered separately, with a strong partnership. However, business-specific context is critical (every company is different) and there are certain advantages (as highlighted) to a combined role.

As a result, I am optimistic and energised about my expanded role as a combined Chief Technology Officer (CTO) and Chief Information Security Officer (CISO).

I recognise the need to establish and empower a strong team, as well as a clear strategy that promotes appropriate autonomy to ensure success. This will be my focus over the first few months!

Microsoft Zero Trust

Tue, 05 Apr 2022 00:00:00 -0500

Earlier today, Microsoft published a customer story focused on my companies journey to Zero Trust.

In the article, Murtaza Nisar (CISO) and I highlight our philosophy and approach to Information Security, including our use of the Microsoft Security Suite, covered under the Microsoft 365 E5 license.

If you would like to learn more about Zero Trust at Microsoft, I would recommend the following links.

We also recorded a live interview, which compliments the article.

The video was recorded after a day of back-to-back meetings, hence I look a little disheveled.

Data to Power Decisions

Tue, 05 Apr 2022 00:00:00 -0500

My colleague Sam Miller recently published an article on LinkedIn, which explores opportunities for data solutions to deliver a foundation that can be used to power decision-making.

The article covers five key areas:

Common Design Pattern and Technologies
Data Products
Decentralization
Frameworks (Accelerators)
Accessibility

I look forward to learning more from Sam and the team over the coming months.

Docker and Apple Silicon

Tue, 15 Mar 2022 00:00:00 -0500

I have been using Docker since 2014, primarily to support web development.

Last year, I tested Docker for Mac on the MacBook Air (M1) and MacBook Pro (M1 Max) both running Apple Silicon, which I documented across four articles.

Docker running on Apple Silicon demonstrated tremendous potential. Unfortunately, the hardware was held back by the software, specifically Docker Desktop for Mac that leverages gRPC-FUSE for file sharing, which resulted in poor I/O performance.

In my testing, the best build time I could achieve running Docker on macOS was 220 seconds, compared to just 9 seconds when building under Docker via a Linux virtual machine (Ubuntu Server ARM64) running on UTM.

The I/O performance limitation is workload-specific, primarily impacting projects with a lot of small files. However, it was disruptive enough, that it forced me away from Docker Desktop for Mac.

Therefore, I have spent the past six months running Docker inside a “right-sized” Linux virtual machine, which unlocked the raw performance of the Apple Silicon by removing the filesystem I/O bottleneck. This approach delivered an unbelievable 24x performance increase over Docker Desktop for Mac.

Docker Desktop v4.6.0

Yesterday, Docker released Docker Desktop for Mac v4.6.0, which included the following new feature:

Docker Desktop 4.6.0 gives macOS users the option of enabling a new experimental file-sharing technology called VirtioFS. During testing VirtioFS has been shown to drastically reduce the time taken to sync changes between the host and VM, leading to substantial performance improvements.

The accompanying blog post states.

“The 4.6 release of Docker Desktop for Mac contains a number of changes that drastically improve file sharing performance for macOS users. Firstly, developers now have the option of using a new experimental file sharing implementation called virtiofs (the current default is gRPC-FUSE). Secondly, improvements have been made to the way that files are synced between the macOS host and Docker VM. During testing with our amazing macOS community of users, we have observed that these changes have reduced the time taken to complete filesystem operations by up to 98%.”

In theory, this new feature directly targets the issue I have been experiencing, therefore I was eager to see how it performed across my common workloads.

Test Setup

As part of my testing, I will be using the 16-inch MacBook Pro (OCT-2021). A summary of the specification can be found below.

16-inch MacBook Pro (OCT-2021)

Apple M1 Max (10-core - 8 Performance / 2 Efficiency)
32-core GPU (10.4 Teraflops)
64GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
macOS 12.3

I plan to use LifeinTECH (this blog) as my test build, which follows the JAMstack web development architecture, using the Jekyll Static Site Generator.

The project includes a range of external dependencies and custom plugins. It is fairly large, including over 4000 individual files and 800MB of data. This is relevant as it applies file sharing I/O pressure.

On the MacBook Pro, I plan to test three scenarios, specifically Docker Desktop for Mac (default), Docker Desktop for Mac (VirtioFS), and Docker running on a Linux virtual machine using UTM.

To enable VirtioFS, update to Docker Desktop for Mac v4.6.0 or higher, select “Preferences > Experimental Features” and check the following configuration options.

Use the new Virtualisation framework
Enable VirtioFS accelerated directory sharing

The screenshot below highlights the configuration.

Once configured, Docker will restart with VirtioFS enabled by default.

Results

The table below highlights my findings. As a point of comparison, I have retained my original test results, which include my AMD Ryzen 3950X (16C/32T) / 64GB RAM custom-built PC running native Docker Desktop (WSL2).

I am pleased to report that the new VirtioFS file-sharing implementation does make a significant difference.

The build time dropped from 220 seconds to 38 seconds, which is also faster than my custom-built PC. It cannot compete with the Linux virtual machine using UTM, but this is essentially Apple Silicon running on native Linux, therefore removing all middleware bottlenecks.

Conclusion

In a perfect world, I would love to see “native” Docker performance under macOS, achieving the same 9 second build times achieved with Linux.

However, 38 seconds is a significant improvement over 220 seconds and it also outperforms my custom-built PC (75 seconds). This is a huge milestone and I commend the team at Docker for this release!

As a result, I believe Docker Desktop for Mac has now become viable for daily use.

Although I will likely retain my Linux virtual machine as a backup, the ability to simply build directly from macOS without any additional pre-work (UTM configuration, Ubuntu Server setup/maintenance, SSH) is highly valuable.

TechPoint MIRA Awards

Mon, 14 Mar 2022 00:00:00 -0500

Update - Unfortunately, we did not win the main prize. However, the team had a great time at the gala and it was great to engage the community. Congratulations to the winners!

LifeinTECH is my personal blog, where I share my thoughts and showcase projects.

However, I do occasionally share some highlights from work, specifically in areas where IT has publically disclosed acomplishments. For example:

The past two years have been unlike anything I have experienced in my career.

In late-2018, we completed an IPO, becoming a standalone company.

Before the IPO, all IT services were consumed from the parent company, supporting thousands of employees and millions of customers across more than 90 Countries.

As a result of the IPO, my team had just over two years to rebuild IT from the ground up, covering the architecture, technology, processes, as well as recruiting the team itself.

This was completed under extreme time and budget constraints, whilst also maintaining business continuity throughout.

During the transition, in 2019, we demonstrated our ambition as a standalone company, by acquiring one of our largest competitors.

Therefore, not only did we need to rebuild IT but also integrate another company simultaneously, which doubled the internal user-based to 15,000.

Finally, these activities were completed under the backdrop of the global pandemic, where travel logistics and supply chains were heavily impacted.

This “perfect storm” of IPO, acquisition and global pandemic feels like a unique, once in a lifetime experience.

Therefore, as an Indiana-based company, I am pleased to share that the team have been nominated and selected as a winner for two TechPoint MIRA Awards.

The TechPoint Mira Awards are Indiana’s largest, best known, and most prestigious technology awards. A total of 276 people, products, companies and places have been honoured with coveted Mira Awards over 23 years, and they have gone on to become the state’s greatest tech successes.

The team have been nominated for two awards:

Large Enterprise of the Year

The Large Enterprise of the Year award recognizes tech companies (including tech-enabled) that are succeeding in the marketplace; investing in innovation; committing to diversity, equity, and inclusion; boosting Indiana’s reputation nationally/internationally; and leveraging their resources and scale to make a positive impact in our community. We especially encourage applications from employers who are advancing diversity, equity and inclusion and/or creating employment and development pathways to underrepresented individuals.

Service Partner of the Year

The Service Partner of the Year award celebrates exceptional work by professional services firms or internal teams who created the year’s best solutions, which are defined as the development of tech products or other professional services for a tech company. Services may be related to IT/cyber, design/branding, marketing/PR, legal/accounting, consulting, and data science.

Full details regarding the awards, including the eligibility, nomination and winners process can be found in the TechPoint MIRA Awards Entry Kit.

Last week, I was interviewed by the executive judging panel, with the winners announced at the annual TechPoint MIRA Awards Gala on 23 April 2022.

Wish us luck!

Swift Playgrounds 4

Tue, 01 Feb 2022 00:00:00 -0600

In December, Apple released Swift Playgrounds 4, which delivers new tools that enable developers to build and submit SwiftUI iPhone and iPad apps directly from Swift Playgrounds.

Previous versions of Swift Playgrounds had been exclusively focused on education, with pre-defined, heavily constrained lessons. Although the previous content was excellent, it failed to provide a path from “basic programming concepts” to building an actual app. For example, the jump from Swift Playgrounds to Xcode is significant, also requiring the developer to have access to a Mac.

With Swift Playgrounds 4, Apple has (for the first time) delivered a development environment that can officially build and submit an app to the Apple App Store without the use of Xcode.

Key new features include:

Swift 5.5, introduced language capabilities for concurrency.
Focuses on SwiftUI apps, but also supports UIKit.
With an Apple Developer Account, developers can submit apps to the App Store.
Supports Swift Package Manager, including import from GitHub.
Includes code completion, similar to Xcode.

As the name suggests, SwiftUI delivers a set of tools and APIs for user interface development, including key features such as GPU acceleration, accessibility options, retina support, widgets, etc. In short, it allows developers to build consistent, great-looking apps across all Apple platforms using Swift.

With this in mind, Swift Playgrounds 4 is still limited regarding the types of apps you can create. For example, it would be difficult to build the next “blockbuster” game. With that said, building on its heritage as an education tool, SwiftUI is a logical next step for Swift Playgrounds, opening the door to new experiences, whilst ensuring an appropriate learning curve.

I had previously used Swift Playgrounds with my son (now six years old), helping him to learn the basic concepts of programming.

Alongside Swift Playgrounds, my son has also been learning via different iPad apps (e.g. Kodable) and is a regular player of Roblox and Minecraft, which (can) incorporate programming concepts.

Therefore, we decided to take these new skills and attempt to build his first app using Swift Playgrounds 4.

As with any project of this type, finding an idea with a defined scope is critical. For example, an app that would appeal to a six-year-old, and could be developed in around 30mins (maximum attention span).

This led me to the excellent tutorial “Build your first SwiftUI app with Swift Playgrounds 4 for iPad” by Paul Hudson. I have embedded the video below, which outlines the process end-to-end.

The app displays a list of personalised photos, which play a custom sound when tapped. It can be developed in around 30 lines of code and includes key concepts such as SwiftUI views, modifiers, use lists, grids, images, buttons, and even the use of a third-party Swift package from GitHub.

My son and I worked together on the app, with him typing/tapping, whilst I provided commentary as to “what” and “why”.

Thanks to the new code completion capabilities of Swift Playgrounds 4, my son was able to start typing (first few letters) and then select from a list to complete the line of code. Now that he is old enough to read, this was an engaging way of coding, as the process of finding the right code snippet was like a mini-game in itself.

All media (photos and sounds) were created and imported into the app directly on the iPad using the included camera and sound recorder apps. This worked well, as it broke up the coding with some fun elements. For example, my son enjoyed making silly sounds, especially when he realised he could make my photo fart.

The entire process took us around 40mins, resulting in a fully functioning app that can be compiled and executed directly from the iPad (no Apple Developer Account required). The full code (minimum viable product) can be found below.

Following the initial release, my son went back and modified the code by repeating the previously learnt steps to add his mum and sister (three years old). This was gratifying, as it demonstrated his understanding and allowed other members of the family to contribute.

Overall, I am impressed with Swift Playgrounds 4. I am excited to see Apple add more advanced capabilities to the app, highlighting a future where end-to-end development could be viable on an iPad. I did notice a few small bugs, but they did not impact the experience and were easily mitigated.

I should also reiterate and commend the excellent tutorial created by Paul Hudson! Great work!

LifeinTECH 2022

Mon, 31 Jan 2022 00:00:00 -0600

I launched my personal blog (LifeinTECH) in 2008.

At the time, I had just started a new job as a Telecoms (Network) engineer. Over the past 14 years, LifeinTECH has followed my journey in technology, providing a location for me to share my thoughts and projects.

As I look back on the 300+ articles (approximately twenty-one per year), It is gratifying to see the trends that have followed my career through multiple engineering roles, to architecture, culminating in my current position as Chief Technology Officer (CTO).

LifeinTECH is a non-profit (no advertisement or tracking) and is not associated with or sponsored by any third parties. This is important, as the primary audience for LifeinTECH is me!

I find writing a powerful way to organise my thoughts and work through complex topics. It also provides a mechanism to help retain (and structure) information (like a journal).

The advantage of having the content public is that others can also (potentially) benefit and/or contribute. It also acts as a nice portfolio, capturing accomplishments and providing insight into passion areas.

Architecture

LifeinTECH follows the JAMstack web development architecture, using the Jekyll Static Site Generator. It is hosted on Netlify, secured using HTTPS, with a certificate provisioned from Let’s Encrypt.

I leverage the Netlify “Starter” plan, which includes 100GB bandwidth and 300 build minutes per month. The advantage of this plan is that it is completely free, whilst still offering modern capabilities, such as Custom Domains, CI/CD, GitHub integration, CDN, Site Previews, Serverless Functions, etc.

The site is configured as a Progressive Web App (PWA), meaning it can be installed on compatible systems. Only the twelve most recent articles, including the “About” and “Ninja Caves” pages, are cached for offline access.

The following Jekyll plugins (Ruby Gems) are leveraged to enable and maintain specific features.

This architecture (JAMStack + PWA) delivers exceptional performance and reliability, whilst also having a very low compute/bandwidth impact (with cached content).

Analytics

As previously stated, LifeinTECH does not target a specific audience and is not associated with or sponsored by any third parties.

There was a brief period in 2009 when I added referral links, which provided an income for a gadget fund. However, I found that this biased my writing and removed the enjoyment of the process. Therefore, these were quickly removed.

As a result, traffic is fairly modest. In January (01-JAN-2022 to 31-JAN-2022), LifeinTECH was viewed by 3,400 unique users, covering 5,220 page views.

Over the years, I have seen peaks (as many as 1000 unique users per day), but I would state the January 2022 traffic flow has become fairly typical.

LifeinTECH also hosts the executable files for my game “Ninja Caves”, as well as the associated Privacy Policy, which is linked to the relevant App Stores (Apple App Store, etc.)

As a result, it is not uncommon for LifeinTECH to pass 50GB of used bandwidth per month. Thankfully, I have never crossed the 100GB limit of the Netlify “Starter” plan.

Over the years, I have used multiple hosts, including Heroku and GitHub Pages, usually gravitating to the best “free” service.

2022 and Beyond!

In 2022, traditional blogging has become less popular, replaced by micro-blogging, vlogging and an infinite number of social networks.

Recognising the goal of LifeinTECH, I have no interest in shifting to a more “modern” approach. I believe there is power in traditional blogging, especially when used as a journal, helping to document, structure and share thoughts/projects.

Therefore, expect to see a steady stream of posts to LifeinTECH, usually biased toward my latest area of interest. Although not the primary goal, I hope the content can help others, and I always welcome input.

Finally, it is important to reinforce that LifeinTECH is my personal blog and in no way reflects the views or plans of my employer.

Kubernetes

Fri, 21 Jan 2022 00:00:00 -0600

Earlier today, Honeypot released an excellent two-part documentary on their YouTube channel, sharing the history of Kubernetes.

Kubernetes (AKA K8s), is an open source system for automating deployment, scaling, and management of containerised workloads. Kubernetes works with Docker, Containerd, and CRI-O and has become the “de facto standard” for cloud-native applications.

I highly recommend watching the documentary, which includes incredible access to Google, Red Hat, and Docker employees, as well as many other influential members of the open source community.

Part one of the documentary focuses on the history of Kubernetes, including the emergence of Docker, Google Borg (Cluster Management) and the infamous “container wars”.

Part two of the documentary describes the growth of Kubernetes, including key adoption milestones and the inevitable market consolidation.

I started my journey with Docker in 2014 and was a big advocate of Heroku following their acquisition by Salesforce.com in 2010, where I worked closely with their team promoting the Twelve-Factor App methodology and refining the developer experience.

I have also spent many years advocating the value of software engineering and open source within traditional businesses, looking to unlock new value through digital business models.

Therefore, it was fascinating to see the team at Google face relatable challenges convincing senior business leaders to sponsor an open source initiative. It highlights that even “cloud native businesses” suffer from bureaucratic inertia, looking to protect legacy/obsolete forms of monetisation.

I commend Honeypot for investing the time to create well researched, high-quality documentaries providing insight into key technologies and the communities that fuel them.

I respect these are niche topics, with a specialist audience, but the impact of technologies such as Kubernetes should not be underestimated, therefore we must document and celebrate these stories.

Nikon D3500 Webcam

Sun, 16 Jan 2022 00:00:00 -0600

Over the past two years, I have spent more time in front of a camera than ever before in my life, video conferencing meetings or virtual events, such as conferences, webinars, podcasts, etc.

I use the excellent Logitech Brio Ultra HD Pro webcam, but even at 4K 30FPS, it lacks the image quality of a Digital Single-Lens Reflex (DSLR) camera.

Therefore, I decided to connect my Nikon D3500 DSLR camera as an optional alternative.

The Nikon D3500 is a popular beginner DLSR, which can be purchased for around £500. Due to some inconvenient software limitations (covered later), the Nikon D3500 is not the best DLSR camera for use as a webcam but is certainly usable in most (non-professional) scenarios.

To get started, you will need to purchase the following hardware, which cost approximately £40.

I selected the Elgato CAM LINK 4k as it offers great quality/performance for the price. However, it is worth noting that it only officially supports Windows and macOS, not Linux.

With the required hardware procured, the setup process is very simple.

Step One: Connect the Elgato CAM LINK 4k and install the Camera Hub software.
Step Two: Insert the Neewer EP-5A for EN-EL14 Power Connector.
Step Three: Turn on the Nikon D3500 and configure the shutter view to 30mins.
Step Four: Swith the Nikon D3500 to “Live View” by clicking the “Lv” button.
Step Five: Connect the HDMI cable to the Elgato CAM LINK 4k and Nikon D3500.
Step Six: Launch the Camera Hub software to confirm it is working.

The camera will now be available to select from video conferencing applications (e.g. Microsoft Teams, Zoom, Cisco WebEx, GoToMeeting, etc.) and streaming software (e.g. OBS Studio, etc.)

You will need a separate microphone, either via a headset or a dedicated device. I use a Blue Yeti connected to the Blue Yeticaster, which includes the Compass Boom Arm and Radius III Custom Shockmount.

With everything configured, you should see a dramatic improvement in video quality over the majority of consumer-grade webcams.

One frustrating software limitation is the Live View timeout, which has a maximum preset of 30mins.

As a result, the camera requires manual user intervention (a button click) once every 30mins to maintain the Live View. The good news is that any button click is acceptable and it does display a warning. Even if you miss the timeout, it does not stop the video feed, it simply goes dark (therefore is quick and easy to recover).

To get the most out of the Nikon D3500, I recommend the following settings.

Set the Live View timeout to 30mins by clicking “Setup Menu” > “Auto Off Timers” > “Custom” > “Live View” = “30m”.
Click the “Exposure Compensation” button, located on the front of the camera to reset the 30mins Live View timeout.
Remove the on-screen overlay by repeatedly clicking the “info” button.
Enable continuous autofocus by selecting mode “P”, switch to “Live View” by clicking the “Lv” button, click the “i” menu button and select “AF-F” (full-time-servo AF).

I don’t use the Nikon D3500 as a webcam every day, as it is less convenient than the Logitech Brio Ultra HD Pro. However, in scenarios where the extra quality is desired, having this setup available is great!

Corporate Separation

Fri, 14 Jan 2022 00:00:00 -0600

Corporate separations don’t happen overnight; they take an incredible number of resources, time, strategic planning, and rapid delivery.

In 2019, my company embarked on the journey to complete a corporate separation in just two years, which was described by many as an impossible mission. However, with an extraordinary team, nothing is impossible!

Our partner World Wide Technology (WWT) recently published a case study, highlighting this amazing accomplishment!

Thanks to WWT for the partnership!

My Setup (Q1 2022)

Fri, 07 Jan 2022 00:00:00 -0600

Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

Home Setup

At home, I switch between three devices (one desktop PC and two notebooks).

My daily driver is a custom-built desktop PC. It is primarily used for productivity, collaboration, software development, photo editing, virtual labs, gaming and game development (GameMaker Studio 2, Godot).

The full specification can be found below:

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
NVIDIA GeForce RTX 3090 FE 24GB GDDR6X
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

The AMD Ryzen 3950X, Samsung 980 Pro and Nvidia GeForce RTX 3090 are premium components, delivering high-performance across a range of workloads (e.g. multi-threading, ray-tracing gaming, machine learning, video editing, etc.)

When remote or travelling, my daily driver is a 16-inch Apple MacBook Pro (OCT-2021). Similar to my custom-build PC, it is primarily used for productivity, collaboration, software development, photo editing, video editing, virtual labs and game development.

The full specification can be found below:

16-inch MacBook Pro (OCT-2021)
Apple M1 Max (10-core - 8 Performance / 2 Efficiency)
32-core GPU (10.4 Teraflops)
16-core Neural Engine
64GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
3x Thunderbolt 4, HDMI, 3.5mm Headphone, SDXC Card Reader, MagSafe 3
16.2-inch Liquid Retina XDR Display (3456x2234 @ 120Hz, 1600nits)

Finally, I use a Framework Laptop running (Fedora) for Linux specific workloads (e.g. Docker). I am a believer in the “right to repair”, therefore eager to support companies promoting this cause.

The full specification can be found below:

Framework Laptop DIY Edition
Intel i7-1280P 4.80GHz (14C/20T)
64GB Crucial DDR4 PC4-25600C22 3200MHz RAM
1TB Western Digital Black SN850 NVMe (7GB/s Read)
Intel Iris Xe Graphics
13.5-inch IPS LCD Display (2256x1504 @ 60Hz)
2x USB4 (USB-C), 1x USB 3.2 G2 (USB-A), 1x HDMI 2.0b

At home, the desktop PC and notebooks connect to a 49-inch Super Ultra-Wide monitor, specifically the Samsung C49RG90.

The full specification of the monitor can be found below.

Model: Samsung C49RG90
Panel Size: 49-inch
Panel Type: VA
Panel Curvature: 1800R
Aspect Ratio: 32:9
Resolution: 5120x1440
Refresh Rate: 120Hz
Variable Refresh Rate: AMD FreeSync 2 (48-120Hz Range)
Response Time: 4ms (GTG)
Colour Accuracy: 125% sRGB, 92% Adobe RGB, 95% DCI-P3
High Dynamic Range: HDR1000
Brightness: 600cd/m2 (Typical), 1000cd/m2 (Peak)

My workspace is kept fairly minimal, the Apple MacBook Pro connects to the monitor directly (DisplayPort v1.4), with peripherals connecting via a UGREEN 5-in-1 USB-C Hub, which simplifies the cable management.

As you can see, for peripherals, I primarily use the Logitech MX Mechanical, Logitech MX Master 3S, Logitech Brio Ultra HD Pro, and Logitech Litra Glow. The keyboard and mouse were selected as they include “easy-switch”, which makes it simple to toggle between multiple Bluetooth devices.

The photo also highlights my DSLR camera and microphone setup. The DSLR camera is a Nikon D3500, connected via an Elgato CAM LINK 4k, which delivers low-latency 2160P at 30fps.

The microphone is a Blue Yeti connected to the Blue Yeticaster, which includes the Compass Boom Arm and Radius III Custom Shockmount. The microphone is primarily used for video conferencing, screencasting, webinars and podcasts.

The photo below provides a closer look at the Samsung C49RG90 monitor.

As highlighted in the photo below, my desktop PC is hidden under the desk, with the case door exposed providing easy access to the components.

Accompanying the desktop PC and monitor is an Original Prusa MINI+, which is an open-source 3D printer.

The desk and matching pedestals are from Three Counties Office Furniture. They are designed for corporate use, therefore are hard-wearing and include integrated cable management.

Finally, I use a Herman Miller Mirra 2 chair, which balances comfort and personalised ergonomics.

Home Server

Alongside my desktop PC and notebooks, I have a small home server, which is connected to my Samsung Q7FN QLED TV. The server manages local and cloud backups, media streaming, and collaboration (video conferencing).

The full specification of the Mac mini can be found below:

Apple Mac mini (NOV-2018)
Intel Core i5-8500B 3.0GHz Base / 4.1GHz Boost (6C/6T)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
256GB PCI-e SSD
Intel UHD Graphics 630
2x 500GB Samsung Portable T5 SSD (USB 3.1 Gen2)

The 1256GB of local SSD storage is split across multiple drives for resilience, with automated backups being completed by Carbon Copy Cloner and Cloud Storage via Google Drive.

To support my minimal (hidden-wire) entertainment setup, the Mac mini is located in a cupboard under the stairs, connected directly to the Samsung One Connect box via HDMI.

The cupboard is a little cluttered, with all the AV equipment (Sony STR-DN1050 AV Receiver, Sony PlayStation 5, Apple Mac mini, Apple TV 4K, Logitech Harmony Hub), but is rarely accessed and offers plenty of room for ventilation.

Work Setup

At work, I use the Apple MacBook Pro and/or Framework Laptop connected to two 27-inch Dell U2718Q IPS HDR 4K monitors via a StarTech Dual-DisplayPort to ThunderBolt 3 Adapter.

The peripherals include a Logitech Craft, Logitech MX Anywhere 3 and Logitech Brio Ultra HD Pro.

The desk is very basic, with a simple dual-monitor stand. Similar to at home, I use a Herman Miller Mirra 2 chair at work.

CTO Confessions Podcast

Sat, 01 Jan 2022 00:00:00 -0600

Last year, I joined IT Labs as a guest speaker on their CTO Confessions podcast.

CTO Confessions is a UK podcast series hosted by IT Labs, a US-based software company focused on Cloud, DevOps, AI, etc.

The podcast series launched in April 2020 and has had some great guest speakers from companies such as Google, Tui, Accenture, easyJet, CloudBees, Revolut, etc.

The 45mins podcast (linked below) covered a wide range of topics, including career, leadership, teamwork, and technology trends.

We also discussed my company journey, specifically our recent transformation that included a major divestiture and acquisition, whilst managing the complexity of the global pandemic.

To access all episodes of the CTO Confessions podcast, I recommend subscribing via one of their channels.

Thanks to the IT Labs teams for the partnership!

LifeinWORK

Mon, 20 Dec 2021 00:00:00 -0600

Although I post frequently about technology topics that relate to my work, I rarely post on my personal blog about specific work accomplishments.

With that said, if you would like to learn more about my work, we recently launched a dedicated page on our corporate website focused on Information Technology.

The hope is to populate this page with information, blog posts, videos, and much more.

Stay tuned!

Dell UEFI Secure Wipe

Sun, 12 Dec 2021 00:00:00 -0600

Earlier this month I needed to return a Dell laptop, which I had been testing via their corporate “try before you buy” scheme.

Historically, I would use a tool such DBAN, GParted or a Linux Live Environment to complete a secure wipe, protecting any sensitive corporate data.

Unfortunately, these traditional tools were designed to erase mechanical Hard Disk Drives (HDD), otherwise known as “spinning rust”. In the era of Solid State Drives (SSD), modern storage controllers continuously optimise data to improve performance, making these traditional tools less effective.

Thankfully, I was pleasantly surprised to learn that Dell laptops manufactured after 2016 include a secure wipe tool accessed via UEFI BIOS, which supports both HDD and SSD. The website “Dell Windows Installation Guide” includes an exellent step-by-step guide.

If the laptop includes an HDD (unlikely in 2021), the secure wipe process defaults to a traditional method, where each bit is zeroed individually. This process is time-consuming (multiple hours) and write intensive, but is effective.

If the laptop includes an SSD, the secure wipe process applies a voltage spike to the drive, which simultaneously zeros all data. This modern method is very fast (almost instant), but highly effective and avoids the intensive writes that can degrade/damage a drive.

Therefore, if you need to secure erase a modern Dell laptop, I would highly recommend using the UEFI BIOS tool (available in SMBIOS Version 2.9 or later). In the context of an SSD, it is fast, write efficient and highly effective.

Printing with Prusa

Sun, 05 Dec 2021 00:00:00 -0600

Over the past month, I have been experimenting with my first 3D printer, specifically an Original Prusa MINI+.

Having recently completed the assembly, the next step was to calibrate the printer and attempt the first print.

I commend Prusa on their excellent documentation and knowledge base, which includes some useful videos. For example, the video below provides an overview of the calibration process, delivered by Josef Prusa.

In addition to the video, I have produced a simple diagram (below) that highlights key 3D printer terminology. The digram is purposely high-level and generic (not specific to the Original Prusa Mini+).

The goal of the calibration process is to ensure the nozzle is set to the appropriate height, allowing for good first layer adhesion. The nozzle can be adjusted in real-time using the “Live Adjust Z” function, however, it is important to ensure it is not lowered to the point that it blocks the plastic from exiting the extruder.

The image below highlights the desired outcome, detailed in the “first layer calibration” knowledge article.

I used Prusament PLA Prusa (Galaxy Black) and the smooth PEI sheet, which I cleaned using an isopropyl alcohol (90%) wipe, but I understand that warm water and a few drops of washing-up liquid works equally well.

It took me several attempts to calibrate the nozzle height, eventually achieving an outcome that closely matched the example. It should be noted that any adjustment of the nozzle height during the calibration process can impact the quality of the print. Therefore, I found it useful to repeat the process until I achieved a clean run from start to finish.

Once complete, I plugged in the included USB drive and selected my first print (a whistle).

PLA (or polylactic acid) is a great plastic material for beginners, as it is vegetable-based (commonly cornstarch) and is fully biodegradable. It also operates at a relatively low temperature, with the Original Prusa Mini+ defaulting to 215 degrees at the nozzle and 60 degrees on the heatbed. These temperatures improve safety and reduce the impact of particle contaminants, specifically volatile organic compounds (VOCs) and ultrafine particles (UFPs).

The print took 33mins to complete, successfully producing a fully functioning whistle.

This early success triggered me to explore the “Prusa Prints” library, which includes thousands of print files. My son (six years old) and I agreed that Pokemon would be a good starting point, recognising there are hundreds of small, relatively simple models.

After a few hours, we had printed a small army of Pokemon, ready to be painted (I understand acrylic paint is the best beginner option for PLA).

I did have a couple of failures, where the print become dislodged from the smooth PEI sheet. Thankfully, this was easily resolved through some additional tuning of the nozzle height. Although the setting is different for every printer, I settled on a nozzle height of -1.500.

Having become comfortable with single file prints, my next challenge was to produce a print that included multiple parts.

Coincidentally, my sister in law was looking for a Golden Snitch from Harry Potter that could be used as a cake topper for my nieces upcoming birthday.

As with any Harry Potter merchandise, online prices are ridiculous (£20+). Therefore, following a brief search, I found a great example on the “Instructables Workshop” website.

I did need to fractionally modify the “Snitch” print file using PrusaSlicer, helping to ensure good first layer adhesion.

The three required prints took approximately 40mins, producing a reasonably impressive result, specifically the intricate detail on the wings.

Finally, throughout the printing process, I did monitor the air quality, which did not show any significant increase, even with the monitor placed directly next to the printer.

With that said, I may look to purchase/build an enclosure and would always recommend printing in a well-ventilated room.

That’s it! My first prints are complete! Over the next few weeks, I hope to increase the complexity of the prints, as well as improve my skills with PrusaSlicer to modify and create print files.

Microsoft Linux?

Thu, 02 Dec 2021 00:00:00 -0600

Microsoft Windows is still the dominant desktop operating system, with over 70% market share. However, the relevance of Windows has shifted dramatically over the past decade, predominantly driven by the growth of web and mobile.

In 2001, Microsoft was best known for Windows, serving as a critical part of their business model and brand. Fast forward to 2021, Windows has been rightly relegated to a bullet point under the “More Personal Computing” business segment.

Outlined below are the three business segments used by Microsoft to communicate with their investors.

Productivity and Business Processes (Office 365, Teams, Dynamics, Power Platform, LinkedIn, etc.)
Intelligent Cloud (Azure, Server Products, GitHub, etc.)
More Personal Computing (Windows, Devices, Gaming, Search, etc.)

These business segments provide insight into how Microsoft prioritises its value and investments, with a clear focus on services, specifically productivity and cloud targeting enterprise businesses.

Even with this prioritisation, Windows remains an important product, installed on more than 1.5 billion devices worldwide.

As highlighted in the 2017 blog post “The Largest Git Repo on the Planet”, the maintenance of Windows continues to be a significant overhead for Microsoft.

For example, the Windows source code includes an estimated 3.5 Million files, stored in a 300GB Git repository. Approximately 4000 engineers work on Windows, producing 1760 daily builds across 440 branches.

This ridiculous complexity and scale likely contribute to some of the challenges experienced by Microsoft when attempting to deliver reliable feature updates for Windows 10, forcing them to drop “Windows as a Service”, moving to a more manageable yearly update schedule.

Commercially, consumer Windows updates are offered for “free”, which combined with the slowing PC market, erodes the return on investment.

Corporate licensing has a similar challenge, knowing that Windows is bundled as part of Microsoft 365.

In short, is there any value in developing and maintaining a proprietary desktop operating system in 2021?

What if Microsoft were to explore an open-source alternative?

Microsoft and Linux

Since Satya Nadella replaced Steve Ballmer as CEO in 2014, Microsoft has demonstrated a greater willingness to support Linux and open-source software.

For example, Visual Studio Code, Edge, Teams, and SQL Server are all available and supported on Linux.

With Edge being based on the open-source Chromium Project, Microsoft web apps are also cross-platform compatible, specifically Office for Web, SharePoint and the Power Platform. Gone are the days where Microsoft web apps require a proprietory browser, extensions or ActiveX controls.

Microsoft has also actively embraced open-source, with “home-grown” examples being .Net Core, TypeScript, Fluent UI, alongside notable contributions to many community projects.

Finally, Microsoft maintains their own Linux Kernel, which is used to power Windows Subsystem for Linux (WSL2). Enabling a Linux kernel within Windows was a significant step towards convergence, making Windows a POSIX compatible development environment and opening the door to native Linux containers (e.g. Docker), etc.

Outside of Microsoft, the community is also contributing to projects that aim to enable Windows software to run on Linux. The most notable are Wine and Proton, which deliver a compatibility layer that allows complex Windows software to run natively on Linux without relying on virtualisation and/or emulation. In recent months, Valve has put its weight behind Proton, setting the ambitious goal of achieving compatibility for 99% of games available via Steam.

Microsoft Linux?

With these examples in mind, I question if the time is right for Microsoft to position Linux as their primary desktop operating system, potentially superseding Windows.

This could be achieved through the delivery of a Microsoft Linux Distribution with a new Desktop Environment that mimics the Windows 11 UI (excluding the legacy UI components).

To the end-user, this new Desktop Environment would look and feel like Windows 11, but behind the scenes, Microsoft would shift from owning/maintaining the full-stack in favour of open-source collaboration.

Over time, I believe this would allow Microsoft to dramatically reduce the overhead associated with maintaining a desktop operating system, as well as drive positive public relations by embracing and investing in the open-source community.

I am not suggesting this approach would be easily achieved, requiring significant upfront investment from Microsoft to deliver a Linux Distribution and Desktop Environment, whilst also contributing to popular community projects to ensure software compatibility, etc.

Outlined below are five areas I believe would require focus to successfully transition to Linux.

Linux Distribution - Deliver/Contribute
Package Management - Contribute
Desktop Envrionment - Deliver
Native Software and Drivers - Deliver
Software Compatibility Layer - Contribute

Ignoring the shareholder and community ramifications, which I suspect would be polarising, the sections below include some thoughts regarding each area.

Linux Distribution

Although Microsoft maintains their own Linux Distribution used for Windows Subsystem for Linux (WSL2), I would recommend they base future development on an established foundation, specifically Debian or Arch Linux.

Knowing the primary audience for Windows is enterprise business, I would recommend Debian, which places a heavy emphasis on stability through their long-term support release model.

Microsoft could explore Fedora as a base, which would be a fun “throw-back” to the IBM PC days, knowing that Red Hat is owned by IBM.

However, I do not believe the “Four Foundations” of Fedora align with the established license-based business model used by Microsoft, which is also still heavily dependant on proprietary software.

With Debian as the base, Microsoft could also exploit this opportunity to make the operating system immutable, as outlined in the article “Immutable OS”,

In short, an immutable operating system makes the root directory (first or top-most directory in a hierarchy) read-only, ensuring it cannot be modified and therefore proactively protecting the core.

This approach would be a significant step forward for security, stability and performance, delivering a foundation that would be immune to “system rot” and accidental/malicious interventions, which have plagued Windows since its inception.

Package Management

Assuming an immutable operating system, Microsoft would need a package management capability that supports dependency isolation (containerisation).

Therefore, I would recommend Microsoft contribute to a project like Flatpak, which would also benefit the wider Linux community.

Although Flatpak has its own set of challenges, I believe Microsoft, with its software expertise, experience and resources could accelerate the maturity of the project, delivering against the promise of a universal software packaging capability with a focus on security and privacy.

Desktop Envrionment

Although Microsoft could build on an established Desktop Envrtionemnt (e.g. GNOME Shell, KDE Plasma, etc.). This is the area I would expect Microsoft to differentiate their Linux experience.

In my opinion, they could take the Windows 11 UI (excluding the legacy UI components) and build a new Desktop Environment from the ground up that is familiar to Windows users, whilst delivering improved performance and consistency.

It would also be an opportunity to explore a modern programming language such as Rust, which emphasises performance, reliability and productivity, helping to future-proof the Desktop Environment.

It is likely Microsoft would want to protect this Desktop Environment, positioning it as a differentiator from other Linux distributions (although it would be great to see them fully embrace open-source).

This approach is not dissimilar to the Microsoft Launcher, which is available for Android.

Native Software and Drivers

As previously stated, Microsoft has come a long way with Linux support, specifically with Visual Studio Code, Edge, Teams, etc.

However, there is still one major gap… Office 365, which includes Word, PowerPoint, Excel, OneNote, Outlook, and OneDrive.

In my opinion, natively porting the Office 365 suite to Linux would have the single largest impact on driving adoption. It would also be a mistake to proceed without support for these applications, knowing that productivity is a key strategic driver for the business.

Outside of first-party software, third-party software has also made great progress over the past few years, as highlighted in the article “Software List 2020”.

A few notable exceptions include Adobe and Serif software, which would require investment and time to port, although could potentially be incentivised and even subsidised by Microsoft to accelerate the outcome.

Regarding drivers, Linux has a surprisingly robust foundation, especially with Intel and AMD hardware and it is arguably more compatible than Windows when it comes to “out of the box” support for peripherals such as printers, scanners, cameras, etc.

However, similar to third-party software, there are notable exceptions. For example, although NVIDIA does provide proprietory drivers for Linux, they are not comparable to the Windows drivers.

With that said, if Microsoft were to position Linux as their primary desktop operating system, I suspect this story would change very quickly.

Compatibility Layer

It is fair to state that with any transition of this nature, some software would be left behind. We have seen this with Apple on several occasions, as they transitioned from PowerPC to x86 and x86 to ARM.

This is where robust a compatibility layer can make a big difference.

In the context of Linux, I would recommend Microsoft contribute and invest in Wine and Proton, helping these projects deliver a greater level of compatibility through native API mapping.

As outlined in the article “Proton - The Trojan Horse”, Proton focuses on graphics APIs, achieved via Direct3D-to-Vulkan translation layers, specifically DXVK for Direct3D 9, 10, 11 and VKD3D for Direct3D 12.

The goal of Proton is to deliver compatibility and (where possible) native performance.

With the recent backing of Valve, Proton becomes a viable (maybe prefered) approach for the development of games on Linux. With official backing and support from Microsoft, I am confident the experience could be improved further, delivering acceptable compatibility and performance for mainstream users.

Outside of Wine and Proton, Microsoft also has Windows 365 and Xbox Cloud Gaming, which enable software (including a full Windows 10/11 desktop) to be streamed over the Internet via a browser.

Although virtualisation and streaming services are not viable for all scenarios, they are becoming increasingly competent and would offer native compatibility, knowing the server-side software could run native code compiled for Windows.

Conclusion

In summary, I recognise that this article does not thoroughly detail the challenges associated with Microsoft attempting to transition to Linux as their primary desktop operating system. It ignores many major barriers and oversimplifies key constraints.

However, I do believe it is an interesting prospect and with the maturity of Linux in 2021, more viable than ever before.

It could even make good business sense for Microsoft, allowing them to focus on their strategic priorities.

Do I think Microsoft will transition to Linux as their primary desktop operating system? Probably not.

Do I think Microsoft should transition to Linux as their primary desktop operating system? Maybe! As long as they do not hurt the FOSS community.

Assuming Microsoft had the appetite to explore this transition, I would highlight the following value proposition.

Microsoft could reduce the overhead associated with the development and maintenance of Windows, allowing them to focus on their strategic priorities, across Productivity, Cloud, Information Security, etc.
Microsoft could deliver a more secure, reliable, and performant desktop operating system that is easier to deploy, support and use. It would also promote transparency and privacy, potentially helping Microsoft avoid future litigation.
Microsoft could use the transition to drive positive public relations by fully embracing and investing in the open-source community, encouraging contributions from diverse talent around the world.

Regarding the return on investment, which is likely what most shareholders care about.

If we take the 4000 engineers working on Windows and assume an average wage of $50,000 per year, that equates to a staggering $200,000,000 per year (without considering other dependencies).

If Microsoft could re-purpose a percentage of this investment into their strategic priorities, they would have a compelling business case.

Finally, regarding a potential go-to-market strategy, I would recommend Microsoft perform this transition as a true open-source project, where the development is transparent, promoting input from the community.

With this in mind, it would make sense to deliver it under a separate brand (maybe EdgeOS) to help differentiate it from the NT-based versions of Windows. This approach would allow the Microsoft Linux Distribution to co-exist with Windows until it is ready for “prime time”.

This concludes my thought experiment (for now). I do not claim to have the required expertise, experience or situational awareness to fully comprehend this topic. Therefore, please consider it a collection of thoughts, which can be fun to explore.

Kontent Horizons

Thu, 25 Nov 2021 00:00:00 -0600

Earlier today, Kentico Software published the recordings from their recent Kontent Horizons 2021 conference.

Kentico is a powerful headless CMS, similar to Contentful and Prismic.

I am a huge advocate of the headless CMS architecture pattern, as it delivers a clear separation of concerns (content/code) following an API-first paradigm, that promotes cloud native applications (Twelve-Factor App) and unlocks multi-channel marketing.

A common failing of this architecture pattern is the marketer experience, which can be disjointed (not fully embedded like a traditional CMS).

Kentico aims to close this gap, delivering a cloud-based (SaaS) platform that benefits everyone (marketers and developers), allowing the team to create engaging online experiences that look and feel great via any channel.

My company recently implemented Kentico, therefore took the opportunity to present our journey at Kontent Horizons 2021. The focus of the story is our migration of 100+ websites, facilitated by our in-house developed “Flexible Web Toolkit”.

Jon Stone, Daniel Willis, Harry Fitzgerald, did a great job!

If you have questions or comments, please do not hesitate to contact the team.

Original Prusa MINI+

Wed, 24 Nov 2021 00:00:00 -0600

As highlighted in my previous article, over the past week I have been building the Original Prusa Mini+ kit.

The “kit” version is shipped as individual parts, requiring manual assembly. The Prusa documentation estimates a 5 to 8 hour build time for one person, with 194 steps.

Thankfully, the assembly instructions (linked below) are exceptional and the package includes all the required tools (e.g. Allen keys, TORX key, pliers, wrench, etc.)

This article highlights my build process, with a few photos and commentary.

I started with the Y Carriage assembly (image below), which secures and manoeuvres the heatbed.

I had expected the Y-motor and Y-axis belt installation to present a challenge, but thankfully the MINI-Y-belt-holder includes the required grooves to secure the belt teeth (no fiddly screws required). Additionally, the tension of the belt can be easily manipulated via the MINI-Y-plate-front (a nice design touch).

I did not enjoy the bearing lubrication process, which reminded me of applying thermal paste, where the quantity can impact performance.

Once complete, I started on the Z-axis assembly (image below), which includes the buddy board enclosure and PCB itself. The buddy board enclosure is used to secure the Z-axis to the previously assembled Y Carriage. The 32-bit PCB is custom-built by Prusa, approximately the same size as a Raspberry Pi 4.

Thanks to the larger components, the Z-axis assembly was relatively straightforward. However, I did need to apply a disconcerting amount of force to ensure the steel rods were fully seated. This is a testament to the 3D printer parts, which feel reassuringly dense.

In my opinion, the next phase was the most complex, covering the X-axis, extruder and print head (image below).

With many small and moving parts, I found myself double-checking my work to ensure the instructions had been followed correctly. The printhead is “where the magic happens”, including the hotend, which can reach temperatures of 280C. The installation of the PTFE tube and SuperPINDA sensor requires accuracy, therefore I would recommend allocating additional time for this section.

My build also included the optional filament sensor (£23), which detects the presence and movement of filament, which adds convenience and helps protect against spoiled builds.

Finally, I installed the Heatbed and LCD (image below). The heatbed has a larger surface area and can reach temperatures of 100C, therefore care should be taken when connecting the cables and attaching the cable cover.

The LCD comes pre-assembled, therefore simply needs to be mounted (one screw) and connected to the PCD via a ribbon cable.

With the assembly complete, I powered on the printer and completed the initial setup, which includes a self-test and automatic calibration. Unfortunately, the self-test failed due to an issue with the Z-axis (image below).

After some troubleshooting (Google and Reddit), I identified the issue. I had over-tightened the trapezoidal nut (image below), which restricted the movement of the Z-axis motor.

With the two screw tension reduced, the Z-axis motor kicked into action and the self-test passed successfully!

That’s it! The build took me approximately six hours, split over several days. It is worth re-emphasising the quality of the assembly instructions, which are among the best I have ever used. Therefore, if you are in the market for a Prusa 3D printer, I would have no hesitation in recommending the “kit” version.

Over the next week, I plan to test the printer, starting with some simple print jobs. Stay tuned for the results.

Immutable OS

Fri, 19 Nov 2021 00:00:00 -0600

I am a long-time user of the Linux distribution Fedora. In recent months, I switched to Pop!_OS as my primary Linux distribiton, however, I still keep a close eye on the Fedora Project, specifically their work on immutable operating systems.

The definition of the word immutable is “unchanging over time or unable to be changed.”

In computing, “immutable” is most commonly associated with Blockchain, specifically the decentralised ledger that uses cryptographic hashes to protect the data from alteration.

An immutable operating system aims to achieve a similar goal by making the root directory (first or top-most directory in a hierarchy) read-only, ensuring it cannot be modified.

The purpose is to ensure the host operating system is protected from accidental or malicious modifications, which improves security, stability and performance by guaranteeing the integrity of the core system.

Every application, tool and/or package installed on an immutable operating system executes code isolated from the root directory, providing clear segregation, as well as a highly effective rollback (restoration) capability.

Fedora Silverblue and Fedora Kinoite are great examples of an immutable operating system.

The primary difference between Silverblue and Kinoite is the Desktop Environment, as Silverblue ships with GNOME and Kinoite ships with KDE Plasma.

As a GNOME user, I have the most experience with Fedora Silverblue. The image below highlights my ARM64 installation running on my MacBook Pro M1 Max.

Fedora Silverblue and Fedora Kinoite leverage three core technologies to initialise and maintain the immutable operating system, as well as support application/tool/package management.

OSTree

rpm-ostree is a hybrid image/package system. It combines libostree (AKA OSTree) as a base image format, and accepts RPM on both the client and server-side, sharing code with the dnf project.

OSTree is a system for versioning updates of Linux. It includes a shared library and suite of command-line tools that combines a “git-like” model for committing and downloading bootable filesystem trees, along with a layer for deploying them and managing the bootloader configuration.

OSTree operates in userspace (code that runs outside of the operating system kernel) and supports any Linux file system.

In short, OSTree is used to initialise and maintain the immutable operating system, facilitating system updates, rollbacks, etc.

Flatpak

Flatpak is a universal packaging system for Linux as well as a utility for software development and distribution.

Flatpak aims to deliver an isolated sandbox environment that includes the runtime and bundled libraries to support the execution of the application/tool/package.

This isolated sandbox environment cannot make any changes to the system without explicit permission from the host.

Flatpak provides a mechanism to install/manage an application/tool/package (including GUI applications) on an immutable operating system, without impacting the integrity of the root directory.

FlatHub is a useful resource that maintains a list of available Flatpack compatible applications/tools/packages.

Podman

Podman is a daemonless container engine for developing, managing, and running OCI Containers on Linux.

Critically, Podman containers can be executed in rootless mode, which reinforces the security model defined as part of an immutable operating system.

As a result, podman is considered more secure than Docker as it does not require root access, making it a great choice for running headless/web applications, tools and packages.

Conclusion

In conclusion, an immutable operating system like Fedora Silverblue and Fedora Kionite provide an exciting glimpse into a future where security, reliability and performance are proactively enforced instead of reactively maintained.

This subtle shift in thinking could have a profound impact on how operating systems are built and supported, whilst also ensuring the “out of the box” user experience is protected, avoiding the inevitable “system rot”.

A great example is the upcoming Steam Deck from Valve, which will ship with a new version of SteamOS (based on Arch Linux) that will be an immutable operating system. This makes sense for an “appliance-like” device that is targeting consumers, where guaranteed core software consistency will help simplify support.

In conclusion, with Linux already making progress towards an immutable future and mounting evidence that Apple is doing the same with macOS, I am intrigued to see how this story evolves.

Prusa Research

Thu, 18 Nov 2021 00:00:00 -0600

As a passionate technologist, I aim to stay up to date with emerging technologies, covering Development, Architecture, Hardware and Software, as well as trending buzzwords such as Blockchain, Artificial Intelligence, Quantum Computing, etc.

One area that has eluded me is the rapidly evolving world of 3D printing. I have watched from the sidelines for many years and certainly see the potential in the technology, but have never had the time to get hands-on.

That changes today, with the arrival of an Original Prusa Mini+ kit, which is an open-source fused deposition modelling 3D printer.

Prusa Research was founded in 2012 by Josef Prusa, a Czech hobbyist, maker and inventor who has become one of the most famous names in 3D printing.

In 2021, there are hundreds of consumer 3D printers available to purchase. However, the Prusa origin story resonated with me, avoiding common start-up tropes such as technology incubators and crowdfunding, instead, relying upon the passion of one man and a workshop.

The video below provides a great overview of the Prusa journey, documenting the road to 100,000 3D printers.

Josef Prusa embodies the mantra “LifeinTECH” and has established a company that positively promotes open-source.

The Original Prusa MINI+ felt like the perfect entry point into 3d printing, thanks to the fact that it shares many of the market-leading capabilities offered by the Original Prusa i3 MK3S+, but at a lower price point (£359 vs. £899).

The Original Prusa MINI+ even includes some enhancements over its bigger brother, specifically the colour LCD, 32-bit mainboard, and enhanced I/O options (e.g. Ethernet, Wi-Fi, USB, etc.)

The “kit” version is shipped as components, with a build time of between six to eight hours. Many of the parts are 3D printed, with open-source hardware, meaning the design is made publicly available, allowing anyone to research, modify or distribute the design.

The Original Prusa MINI+ is also available pre-assembled, but I felt the process of building the 3D printer would be a fun project and also help me learn how the hardware works.

The Original Prusa MINI+ includes the following features:

An 18×18×18cm build volume, which is only slightly smaller than the Original Prusa i3 MK3S+ at 25×21×21cm.
Sensorless homing, automatic mesh bed calibration and replaceable nozzles.
2.8-inch, 240×320 LCD screen with 65.000 colours.
Custom-developed 32-bit “Buddy” motherboard with Trinamic 2209 drivers, delivering OctoPrint compatibility.
A magnetic heat bed and removable spring steel sheets, with two variants available (smooth and textured).
SuperPINDA probe features a high-quality sensor for a fully-automatic mesh bed levelling process.
Robust I/O, including USB, Ethernet (RJ45) and an optional Wi-Fi module.
Three thermistors and comes with a high-quality Delta Electronics 150W power supply.

Overall, the Original Prusa MINI+ is a fully-featured 3D printer from a well-established manufacturer with a strong reputation for reliability and safety. The price point (£359) is highly accessible, either for new users or anyone looking to establish a consumer print farm.

I plan to build the printer over the next week, therefore I hope my next update will include a photo of my first successful print!

EVOLVE 21 - Update

Tue, 16 Nov 2021 00:00:00 -0600

Last month, I took part in the EVOLVE 21 webinar series, hosted by Alcor Solutions.

Mike Henry (Executive Director - TechOps) and I discussed our business transformation journey that included a major divestiture and acquisition.

The session is now available online.

Thanks to the Alcor team for allowing us to share our experience, specifically Ziying “Magic” Tan and Pam Morris for facilitating.

The Linux Challenge

Tue, 09 Nov 2021 00:00:00 -0600

The Linux community has been patiently waiting for the first Linus Tech Tips (LTT) Linux Challenge video to drop.

The Linux Challenge is a five-part series where Linus and Luke have agreed to commit to Linux as their primary gaming system for one month, documenting their experience.

Part One: Select a Linux distribution, install Linux and play a game.
Part Two: Game streaming setup (e.g. capture, microphones, levels, etc.)
Part Three: Install and run a game from as many game launchers as possible.
Part Four: Complete non-gaming tasks, including productivity, collaboration, etc.
Part Five: Conclusion and final decision to remain on Linux

If either Linus or Luke “fail” and are forced to go back to Windows, they must complete a forfeit. Knowing that Luke has previous experience with Linux, I suspect he has a higher chance of success.

As highlighted by the Steam Hardware and Software Survey, Linux has approximately 1% market share, demonstrating a steady (but minor) increase over time.

LTT is the largest tech channel on YouTube by video views and the eleventh largest by subscribers (with more than 14 Million). Therefore, any video published by LTT has the potential to influence a massive audience.

Part one of the Linux Challenge is now available (video embedded below).

Without question, the most controversial part of the video was the choice of Linux distribution (Pop!_OS) and the error encountered by Linus at 09:57, which resulted in him removing essential packages required by the operating system.

In June, I switched from Fedora to Pop!_OS as my primary Linux distribution and have written several articles describing my rationale and setup.

One of the reasons I like Pop!_OS is the “out of the box” experience, delivering what I consider to be the closest equivalent to the Apple “it just works” mantra.

Therefore, I watched in horror, as Linus essentially “killed” his system whilst simply trying to install the popular game launcher Steam.

To understand what happened, the image below provides a closer look at the warning message.

The warning clearly states that the following packages will be removed.

pop-desktop
gnome-control-center
pop-gnome-initial-setup
chrome-gnome-shell
gdm3
xorg3
gstreamer1.0-vaapi

Anyone with Linux experience will immediately identify these packages as critical, knowing they enable the Pop!_OS Desktop Environment (COSMIC, GNOME, Xorg), otherwise known as the Graphical User Interface (GUI).

Therefore, I suspect this warning and the resulting outcome, will generate a lot of discussion amongst the Linux community. Was this a user error or is Linux to blame?

In my opinion, it is a bit of both but ultimately highlights that even a competent technologist must have a basic understanding of Linux to be successful.

The issue originated from a bug, which has subsequently been fixed. This bug would have been resolved if Linus would have had an internet connection enabled during the installation of Pop!_OS or completed a system update post-installation.

In my opinion, a system update is best practice with any operating system (Linux, Windows, macOS) following a fresh installation, as you can almost always guarantee there will be updates and bug fixes.

With that said, as a new Linux user, a simple application installation should never “break” the system.

In fairness, when Linus tried to install Steam from the Pop!_Shop, it simply failed, successfully protecting the system.

However, the error message was essentially useless for a new Linux user, not providing any valuable path to resolution. For example, it could have prompted the user to update their system and try again (which in this case, would have resolved the issue.)

When Linus switched to the Terminal, the installation of Steam generated the same error as the Pop!_Shop, but also provided a (bad) path forward.

Knowing the ramifications of this action, Linux forces the user to manually type the phrase “Yes, do as I say!”. This additional step is designed to warn the user that the following actions could result in unforeseen consequences.

Unfortunately, like the majority of users, Linus did not stop to consider the warning message, instead deciding to proceed regardless (we have all done it).

At which point, Linux executed the exact actions as requested by the user, including the removal of the Pop!_OS Desktop Environment.

Therefore, I feel the “blame” is split between the user and Linux, knowing that both could have acted differently to prevent this outcome.

Overall, the biggest disappointment is that this experience triggered Linus to switch his Linux distribution from Pop!_OS to Manjaro. Although I also like Manjaro (as highlighted in the article Linux Distributions), I do believe the “new user” experience is better on Pop!_OS.

I also believe Linus would be better served by a Debian-based Linux distribution, instead of Arch Linux, which in my experience can have a steeper learning curve.

As a fan of Linux and Pop!_OS, I am not disheartened by this outcome. Linux still has a long way to go before it is ready for general consumers and I hope that this series of honest videos will help shine a light on areas that need improvement.

It will be interesting to see how the Linux community (and System76) respond (get ready for a lot of memes) and I look forward to watching the rest of the series.

Update (10-NOV-2021)

System76 have responded to the issue Linus experienced with Pop!_OS.

“For some reason, an i386 version of a package was never published on Launchpad. Steam being an i386 package when trying to install it, it had to downgrade that package to the Ubuntu version to resolve dependencies, which removed Pop!_OS packages.”

As I had hoped, the issue has also triggered System76 to patch APT, preventing users from accidentally breaking their system via the Terminal. As with all things Linux, this protection can be overwritten but requires a specific user intervention.

Docker Performance on M1

Wed, 03 Nov 2021 00:00:00 -0500

Update - On 14-MAR-2022, Docker released Docker Desktop for Mac v4.6.0, which included a new experimental file sharing implementation called virtiofs. Please refer to my follow-up article “Docker and Apple Silicon” for the latest results.

With the arrivial of the 2021 MacBook Pro, I have been testing my common workloads.

I have been using Docker since 2014, primarily to support web development. Last year, I tested Docker for Mac on the MacBook Air running Apple Silicon, which I documented across three articles.

The performance of Docker running on the MacBook Air was good, relative to Intel equivalent Mac. It was clear that the Apple M1 System on a Chip (SoC) had tremendous potential, but was held back by the software. For example, in my testing, Docker Desktop for Mac was consistently outperformed by its x86 Linux and Windows (WSL2) counterparts.

In this article, I plan to take a second look at Docker Desktop for Mac. I hope that a year of additional development and the more powerful Apple M1 Max SoC will deliver tangible improvements.

Test Setup

As part of my testing, I will be using the 16-inch MacBook Pro (OCT-2021), 16-inch MacBook Pro (NOV-2019), Dell XPS 17 (9710) and a custom-built desktop PC. A summary of the specification for each system can be found below.

16-inch MacBook Pro (OCT-2021)

Apple M1 Max (10-core - 8 Performance / 2 Efficiency)
32-core GPU (10.4 Teraflops)
32GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
macOS 12.0.1

16-inch MacBook Pro (NOV-2019)

Intel Core i9-9980HK 2.4GHz Base / 5.0GHz Boost (8C/16T)
AMD Radeon Pro 5500M 8GB GDDR6
32GB 2666MHz DDR4 RAM
1TB PCI-e SSD
macOS 11.6

Dell XPS 17 (9710)

Intel Core i9-11900HK 2.6GHz Base / 5.0GHz Boost (8C/16T)
NVIDIA GeForce RTX 3060 Ti 6GB GDDR6 (70W)
64GB DDR4 3200MHz RAM
1TB M.2 PCI-e NVM-e SSD
Pop!_OS 21.04 (Linux 5.13)

Custom-Built Desktop

AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
NVIDIA GeForce RTX 3090 FE 24GB GDDR6X
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
Windows 11 (WSL2)

I plan to use LifeinTECH (this blog) as my test build, which follows the JAMstack web development architecture, using the Jekyll Static Site Generator.

On the MacBook Pro, I plan to test two scenarios, specifically Docker Desktop for Mac and Docker running on a Linux virtual machine using UTM.

Docker Desktop for Mac

Docker Desktop for Mac includes multiple configuration options that can directly impact performance.

Under “Docker Desktop > Preferences > Resources > Advanced”, I configured the following settings.

CPUs: By default, Docker Desktop is set to use half the number of CPUs available on the host machine. However, Docker Desktop does not detect the M1 Max correctly, therefore I manually configured 8 CPUs. At this time, there is a software bug that prevents configuring all 10 CPUs.
Memory: By default, Docker Desktop is set to use 2GB runtime memory, allocated from the total available memory on the Mac. This is adequate for my test build but can be adjusted as required. It should be noted that this memory is allocated to Docker Desktop regardless of a running container.
Swap: I retained the default swap of 1GB.

Under “Docker Desktop > Preferences > Resources > File Sharing”, I configured the following settings.

File Sharing: One of the biggest bottlenecks associated with Docker Desktop for Mac is file sharing I/O performance. Therefore, it is important to only share the directories required by the container. For example, the specific “development” folder.

Finally, under “Docker Desktop > Preferences > Experimental Features”, I checked the setting “Use the New Virtualisation Framework”, which enables the new Apple Hypervisor framework that should deliver better performance.

Ubuntu Server on UTM

I installed Ubuntu Server (ARM64) on UTM, provisioned with 4 Cores, 8GB Memory and NVMe Storage.

The Ubuntu Desktop Environment was installed, with the Docker Desktop and Docker Compose packages. The full setup can be found in the article “Virtualisation on M1”.

Results

The table below outlines the results of my testing.

Performance across Linux (Dell XPS 17) and Windows (Custom-Built PC) provides a good baseline as they both build against a native Linux kernel.

The results from the 16-inch MacBook Pro (OCT-2021) are inconsistent depending on the specific configuration. For example, when using Docker Desktop for Mac, it is critical that an ARM64 image is selected and the Apple Hypervisor is enabled to achieve acceptable performance.

In this scenario, the Apple M1 Max performs as well as the 16-inch MacBook Pro (NOV-2019), with an Intel Core i9-9980HK processor. Interestingly, the 16-inch MacBook Pro (NOV-2019) performance degraded with the Apple Hypervisor enabled.

The most exciting result was the Ubuntu Server (ARM64) running on UTM, which achieved a blistering 9 second build time. This result is the fastest I have seen on any system, demonstrating the raw power of the Apple M1 Max when software limitations are removed.

For example, the testing indicates that file system I/O is the limiting factor when using Docker Desktop for Mac, especially with large, multi-file projects. This is not an issue in scenarios where the native Linux kernel filesystem can be accessed directly.

Conclusion

I remain conflicted regarding the performance of Docker for Mac, specifically with the 16-inch MacBook Pro (OCT-2021) running the Apple M1 Max SoC.

If it is not possible to run a Linux host directly, I prefer the implementation of Docker for Windows, which takes advantage of WSL2. In my opinion, this is a great compromise, allowing the developer to use Windows software, with Linux performance.

Regarding macOS, Docker Desktop feels constrained, forced to use middleware that is not always performant (e.g. File Sharing). As a result, I will likely retain a “right-sized” Ubuntu Server running as a virtual machine on UTM.

This approach unlocks the raw performance of the Apple M1 Max SoC by removing the filesystem I/O bottleneck, delivering an unbelievable 24x performance increase over Docker Desktop for Mac.

Finally, it is worth noting that these results are workload-specific, therefore the highlighted filesystem I/O bottleneck may not always be an issue.

Foldable Future - Update

Tue, 02 Nov 2021 00:00:00 -0500

Over the past year, I have been using the Samsung Galaxy Z Fold2 5G as my primary smartphone.

As a reminder, the Fold2 includes a 7.6-inch 120Hz AMOLED display that folds internally, supported by a complex mechanical hinge. The full specification of the device can be found below:

Samsung Galaxy Z Fold2 5G
Qualcomm SM8250 Snapdragon 865+ (Octa-Core)
6.2-inch HD+ AMOLED Display (2260x816 @ 60Hz)
7.6-inch QXGA+ HDR10+ AMOLED Display (2208x1768 @ 120Hz)
12GB RAM
256GB Storage (UFS 3.1)
2x 10MP Selfie Cameras
3x 12MP Cameras (Ultra Wide, Wide-Angle, Telephoto)
4500mAh Dual Battery
5G, LTE, Wi-Fi 6, Bluetooth v5.0
Finger-Print Reader and Face Unlock
USB-C and Wireless Charging

Although over a year old (and superseded by the Fold3), the specification of the Fold2 is still very impressive. However, this specification also commanded a premium price point, starting at £1799.

This article will act as a short update regarding my experience using the Fold2 throughout 2020/21, reflecting on the durability of the hardware.

Regarding my usage pattern, the Fold2 was my daily driver (no other smartphone), commonly stored in my trouser pocket. I did not use a protective case or skin and removed the internal display protector, as I did not like the “plastic” feel.

Although much of the year was in lockdown, the Fold2 was exposed to the elements, specifically gym, running and multiple trips to the beach. I would also regularly join video conferencing calls whilst walking, often in adverse weather conditions (wind, rain, etc.)

In short, the Fold2 was treated with respect, but not coddled.

Infamously, the first generation Fold, released in 2019, was an unprecedented disaster, due to a fundamental design flaw that resulted in an exceptionally high failure rate.

I am pleased to report that I have had no such issue with the Fold2. I must have unfolded the Fold2 thousands of times over the past year and the hinge feels as good today as it did in October 2020.

The hinge as always felt very robust, with enough pressure to hold the two sides of the display at any angle. I had feared the hinge would weaken over time, potentially becoming loose, but there are no obvious signs of degradation.

This is an impressive accomplishment, especially considering the complexity of the hinge and knowing the challenges faced by the previous generation.

The front display, which is protected by Corning Gorilla Glass Victus, shows signs of micro-scratches, but in line with my experience of other smartphones (e.g. iPhone, etc.)

The internal display is more interesting, as it does not have the same level of protection when in use. Thankfully, the display itself has demonstrated great resilience. My only criticism is a series of minor scratches along the fold. It is unclear if these scratches have formed because of the folding mechanism itself or (more likely) the fold continues to be an exposed area (like with the first generation product), which can easily collect particles, such as dust, dirt, sand, etc.

I can feel these scratches when running my finger over them, however, when viewing full-screen content, they essentially fade into the background.

Both displays continue to work as designed, with no loss of touch sensitivity, etc.

This brings me to “the big drop”. Unfortunately, last month I missed my pocket and dropped the Fold2 on a stone floor. As soon as it landed I knew I was in trouble.

Surprisingly, the front and internal display survived! However, the rear glass was not so lucky.

This outcome was not a complete surprise, recognising the rear is protected by Corning Gorilla Glass 6 (not Victus). I would speculate that any glass-backed smartphone would have fallen to the same fate.

Thankfully, the Fold2 itself continues to operate as designed, with no obvious impact on core functionality (touch sensitivity, cameras, etc.)

With that said, I suspect the repair cost would be high, knowing that the repairability of modern smartphones is rarely a priority. Therefore, I might get a dbrand skin to help hide the damage.

Overall, I remain impressed with the Fold2, which has served me admirably over the past year. At no point have I been concerned about the build quality, which has proven to be as robust as any other smartphone I have used.

Considering the complexity of the hinge and folding display, this is a remarkable achievement and a great “real world” demonstration that folding displays are viable for everyday use.

This article has focused on the hardware durability, but it is also worth noting that the software has received steady updates from Samsung.

When I first started using the device I was concerned that most apps would not be optimised for the larger display. Although this is true for some, the majority work very well, arguably offering a better experience than traditional smartphones.

This is especially true for content consumption and multi-tasking, where the Fold2 is in a league of its own.

For example, I would regularly join Microsoft Teams meetings, where one side of the display was populated by the participant videos and/or presentation, with the other available for comments or notes. With picture-in-picture (image below), multi-tasking can be taken to the next level, with three distinct “content windows” displayed simultaneously from a single app.

This level of multi-tasking is simply not viable on traditional smartphones, acting as a real differentiator for business users and/or road warriors.

In conclusion, as highlighted by my previous review, I believe the Fold2 is the “real deal”, delivering on the promise of a foldable product that could replace a traditional smartphone and a small tablet.

After “battle testing” the Fold2 for over a year, I remain convinced that foldable smartphones are the future, something I expect all the mainstream providers to produce, likely starting with the Google Android ecosystem (maybe a future Pixel product).

Virtualisation on M1

Fri, 29 Oct 2021 00:00:00 -0500

With the arrivial of the 2021 MacBook Pro, I have been testing client-side virtualisation on Apple Silicon.

I use virtualisation frequently for testing and to run specialist workloads. Historically, these virtual machines leveraged the x86-64 instruction set, which is native to Intel and AMD processors. However, with the shift to Apple Silicon, the AArch64 (ARM64) instruction set is required to achieve native/optimal performance.

The AArch64 ecosystem is less mature than x86-64 for “desktop” computing, reducing the number of compatible guest operating systems.

Thankfully, the main bases are covered, including Fedora, Ubuntu (Debian-based) and Arch Linux, as well as Microsoft Windows 10/11.

On a Linux or Windows host, I use VMware Workstation for client-side virtualisation, mainly because I have years of experience working with VMware products (e.g. ESXi, etc.).

However, VMware is a little behind with their Apple Silicon implementation, therefore I have selected UTM, which is free and open-source software, but also offered via the Mac App Store (£8.99) if you would like to support the project.

Similar to VMware Workstation (and Parallels), UTM allows you to securely virtualise and emulate guest operating systems on your Mac.

What makes UTM unique is that it employs the Apple Hypervisor virtualization framework to run AArch64 operating systems on Apple Silicon at near-native speeds, as well as lower performance emulation for x86/x64 operating systems.

Behind the scenes, UTM leverages QEMU, which is a mature, free and open-source emulation software that is actively maintained by a passionate community. UTM takes the best parts of QEMU, providing a user-friendly GUI for simple configuration and maintenance.

This architecture makes UTM very versatile, perfect for developers that what to emulate different physical architecture, such as ARM32, MIPS, PPC, and RISC-V.

The only watch-out relates to GPU hardware acceleration, which is not currently supported across virtualisation or emulation. Therefore, do not expect to use UTM for running graphics-intensive software or playing modern games.

Linux Virtual Machine Installation

The following screenshots highlight the installation process for Linux (Ubuntu Server), including the setup of a Desktop Environment (GUI).

Launch UTM and select to create a new Virtual Machine. Provide some basic information (metadata) about the virtual machine, including the name, style and icon.

Under the “System” tab, select the “ARM64 (AArch64)” architecture and allocated the desired amount of memory. For a local, single-user server I normally allocate 8GB (8192MB).

Select “Show Advanced Settings” and specify the desired number of CPU cores. I found that leaving this setting “blank” resulted in a boot error.

Under the “Drives” tab, create two drives. The first will serve as the system drive (NVMe), while the second is used for the installation media (ISO). The allocated drive space is not provisioned upfront, therefore it is safe to specify a reasonable drive size (e.g. 60GB).

Once configured, the virtual machine will be accessible from the UTM Library. At which point. select the CD/DVD drive and attach the desired installation media (ISO).

Start the virtual machine. If everything is configured correctly, you should be presented with the classic Linux GNU GRUB menu, allowing you to select “Ubuntu Server”.

This will launch the Ubuntu Server installation wizard. Use the “Cursor Keys” to navigate, “Space” to check options and “Enter” to select.

The majority of the options can be left as default. However, a few require manual intervention. For example, the profile setup.

You may also want to specify any required features (e.g. Docker) as part of the installation, however, these features can also be added later.

Once complete, the wizard will prompt you to reboot. Instead, close the virtual machine and eject the installation media (ISO).

Start the virtual machine and you will be prompted to log in with the previously defined profile information.

As this is a server product, Ubuntu Server does not include a Desktop Environment (GUI). This is perfectly fine for most server workflows, however, if you would like the Ubuntu GNOME Desktop Environment, run the following commands:

sudo apt-get update

sudo apt-get install tasksel

sudo tasksel install ubuntu-desktop

These commands will update the package information from configured sources and install the Ubuntu GNOME Desktop Environment. The process will take several minutes, but once complete, the subsequent reboot will display the standard Ubuntu GNOME Desktop Environment login screen.

If the installation process fails, simply re-run the ‘install ubuntu-desktop’ command, which should pick up at the point of the failure.

Finally, it is worth installing the SPICE Guest Tools and Drivers, which allows for advanced display setting to be enabled, such as “Retina Mode”, etc.

sudo apt-get install spice-vdagent spice-webdavd

That’s it! Ubuntu Server, with the Ubuntu GNOME Desktop Environment, is now ready for use.

PowerBook Reborn

Wed, 27 Oct 2021 00:00:00 -0500

I purchased my first Apple notebook in 2005, the legendary Aluminium PowerBook G4. This was not my first Mac, but all previous interactions were via second-hand desktops that I collected due to my obsession with technology (not because I had a specific use case).

The PowerBook G4 changed everything! It was my first significant laptop purchase and marked the point I switched from Windows to OS X as my primary operating system.

I still used Windows and Linux regularly (including a year as a Windows/Linux/Unix administrator), but OS X took me through university, including the delivery of my dissertation and associated software development project (which can be found on GitHub).

The (blurry) photo below is from my flat in 2008, proudly featuring the PowerBook G4, Airport Extreme and even the original iPhone (imported from the US).

The specification of the PowerBook G4 was modest by today’s standards, but included many “pro” features not found on Windows laptops at the time (high definition display, FireWire, etc.)

PowerBook G4 Rev E (PowerBook5,8)
PowerPC 7447a (G4) 1.67GHz (1C/1T)
512MB DDR SDRAM (PC2-4200)
80GB 5400RPM HDD
ATi Radeon 9700 128MB DDR SDRAM (4x AGP)
2x USB 2.0, FireWire 400, FireWire 800
Analog/Optical Digital Audio
15-inch TFT Display (1440x960)

The PowerBook G4 served me well for over four years, at which point the Apple transition from PowerPC to Intel was well underway, prompting me to purchase a MacBook Pro.

For fifteen years, the 15/16-inch MacBook Pro was my notebook model of choice, acting as my daily driver at home and for work. Due to my love of technology, I also had Windows and Linux systems, but the MacBook Pro went everywhere with me.

In 2016, Apple released a major redesign of the MacBook Pro, removing many features used by the professional community. Instead, Apple prioritised aesthetics with a slimline design that felt more like an oversized ultrabook.

Although the notebook itself was still among the “best” available, with Apple’s market-leading build quality, the shift towards a design-first philosophy, frustrated the professional community, myself included.

As a result, like many in the community, I decided to switch from the MacBook Pro to Windows/Linux laptop as my daily driver. It took me a while to find the right fit, working my way through three devices (which were all very good).

For the past year, I have used a combination of the Razer Blade 15 Advanced (Windows 10/11) and Dell XPS 17 (Linux - Fedora and Pop!_OS. Both of these laptops deliver exceptional performance, with unique features targetting the professional community, specifically:

High-Performance CPU with Vapor Chamber Cooling
High-Performance GPU with Vapor Chamber Cooling
High-Performance Memory (up to 64GB)
High-Performance Storage (NVMe SSD up to 8TB)
Ports (Thunderbolt, USB-C, USB-A, SD-Card Reader, HDMI, DisplayPort)
High-Performance Power Delivery (140W+)
High Resolution, High Refresh Rate, Touch Display
Professional Orientated Keyboard
Biometric Security (Fingerprint Reader and/or Face Recognition)

In short, the Razer Blade 15 Advanced and Dell XPS 17 consistently outperformed the equivalent MacBook Pro, usually at a significantly lower cost (greater than £1000).

Although these laptops have served me well over the past few years, I found myself imagining a future where Apple returned to the “glory days” of the PowerBook, with a product that prioritised professional workloads.

On the 18th of October, Apple announced the new 2021 MacBook Pro, running Apple Silicon (M1 Pro and M1 Max).

I can honestly state it is the most excited I have been about an Apple event in years and as soon as Apple revealed the design and specification of the 2021 MacBook Pro, I knew my fate was sealed.

Yesterday, my new 16-inch MacBook Pro arrived!

I selected the 16-inch SKU, equipped with the M1 Max System on a Chip (SoC). The full specification can be found below.

16-inch MacBook Pro
Apple M1 Max CPU (10-core - 8 Performance / 2 Efficiency)
32-core GPU (10.4 Teraflops)
16-core Neural Engine
32GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
3x Thunderbolt 4, HDMI, 3.5mm Headphone, SDXC Card Reader, MagSafe 3
16.2-inch Liquid Retina XDR Display (3456x2234 @ 120Hz, 1600nits)

Considering my workloads, I configured the highest performance CPU/GPU option. I did not upgrade the memory or storage, which I felt were not worth the additional cost for my usage.

To validate this thinking, I launched the following applications simultaneously, which would represent an extreme scenario (for me).

Microsoft Teams (Video Conference)
Microsoft Word
Microsoft PowerPoint
UTM (Ubuntu Server - 4 Core / 8GB)
Docker Desktop for Mac (8 Cores / 2GB)
GameMaker Studio 2 Game Engine and YoYo Runner (2D Game)
Godot Game Engine (3D Game)
Visual Studio Code
Google Chrome (Seven Tabs)
Terminal
Finder Preview (PDF)
Activity Monitor

It is worth re-emphasising that this test included a Linux Virtual Machine (Ubuntu Server) and two game engines (GameMaker Studio 2 and Godot), both running compiled games.

I am pleased to report that the Mac remained responsive throughout the test, consuming an average of 23.09GB physical memory and 0GB swap. At peak, I have seen up to 26GB of physical memory used.

Some workloads will apply more memory pressure, for example, 3D modelling, video editing, video effects, etc. However, I am fortunate to have a custom-built desktop PC with 64GB memory to support these memory-intensive workloads.

This article is not a review of the 2021 MacBook Pro, for that, I would reference Anandtech, Dave2D, etc.

Instead, I wanted to highlight my first 48hours with the notebook, which I consider the spiritual successor of the PowerBook G4.

Interestingly, the aesthetics even resemble the PowerBook G4, with its “boxy” design.

It goes without saying, similar to the Apple M1 SoC, the Apple M1 Max SoC delivers ridiculous performance and is a serious warning shot to Intel/AMD/NVIDIA, with a “power to performance” ratio that is not only market-leading but arguably a generational leap over the competition.

As a single reference point, the Apple M1 Max SoC achieved a GeekBench score of 1772 (single-core) and 12688 (multi-core), which destroys my 2019 16-inch MacBook Pro (Intel i9-9980HK), which scored 1032 (single-core) and 6583 (multi-core).

Thanks to the power efficiency of the SoC, this performance is also sustained even when operating on battery, which is not the case with high-performance Windows laptops, often resulting in a 75% performance reduction.

Combined with the advertised 21 hours battery life (realistically 10 hours for mixed workloads), the MacBook Pro is a productivity monster.

Regarding the design itself, I consider the 2021 MacBook Pro close to perfect. Apple has essentially walked back every controversial decision they made in 2016, specifically:

Removal of the Touch Bar, replaced with full-height function keys.
MagSafe 3 delivering 140W of power, charging to 50% in 30mins.
HDMI and SD Card Reader.

This is the MacBook Pro design we should have received in 2016, especially considering this chassis also delivers a larger thermal envelope, which would have suited the thermally inefficient Intel processor. The image below is a stacked 2016, 2019 and 2021 MacBook Pro.

Thankfully, Apple has also retained (and even improved) the keyboard, which has clicky, tactile, responsive keys. This is a long way from the disastrous “butterfly switch” keyboard introduced in 2016.

If that is not enough, the 2021 MacBook Pro also includes a high resolution (3456x2234), high refresh rate (120Hz) display, high-performance storage (7.4GB/s Read) and a thermal system that is so efficient it rarely engages the fans.

In conclusion, I am pleased to reinstate the 16-inch MacBook Pro as my primary notebook! Apple has finally listened to their customers and delivered a product that thoroughly deserves the designation of “Pro”.

EVOLVE 21

Mon, 25 Oct 2021 00:00:00 -0500

Update - The EVOLVE 21 webinar is now available to view online.

On the 29th of October I will be speaking as part of the EVOLVE 21 webinar series, hosted by Alcor Solutions.

Founded in 2008, Alcor Solutions specialise in cloud advisory and implementation services, supporting customers globally across multiple industry verticals. Their expertise includes Automation, IT Service Management, IT Asset Management, Human Resource Automation and Data Analytics.

EVOLVE 21 is a webinar series focused on insights from global thought leaders and industry experts, covering digital transformation and the challenges associated with technology adoption across different verticals. Each webinar includes an interactive section, where the global audience can contribute and ask questions.

My colleague, Mike Henry (Executive Director - TechOps) and I will discuss “Digital Transformation: Large Scale Change”, covering our business transformation journey that included a major divestiture and acquisition.

The webinar is scheduled for Friday, 29-OCT-2021 at 08:00 PDT / 15:00 GMT. It is free to register, therefore I hope to see you there!

vscode.dev

Thu, 21 Oct 2021 00:00:00 -0500

Yesterday, Microsoft released vscode.dev, which is a lightweight version of Visal Studio Code that runs in the browser.

This is not the first time Visual Studio Code has been made available as a browser-based application, however, previous examples have required a dedicated server or a paid subscription (GitHub Codespaces).

vscode.dev is different as it follows a SaaS model (no dedicated server) and is available for free.

Knowing that Visual Studio Code is based on the cross-platform Electron framework, you may be questioning why anyone would need a browser-based version. I can think of two specific examples.

In a corporate/education environment, end-user computing can be restricted through policy, sometimes removing the ability to install software development tools locally. Ironically, these restrictions can even be applied to software developers, inhibiting their ability to work. For example, I have seen scenarios where corporate/education entities require their software development teams to use jump boxes and/or virtual desktop infrastructure, which directly impacts the developer experience and productivity.
Software development on low-powered systems or platforms with inherent software restrictions. For example, Google Chromebook or the Apple iPad. Although there are alternative code editors for these platforms, they rarely compete with Visual Studio Code, which at this point has become the standard.

I am most interested in the second example, specifically software development from the iPad. Over the years, I have challenged myself to use the iPad as a daily driver, specifically in the context of work. Unfortunately, software development has always been a challenge, requiring elaborate workarounds (using a Raspberry Pi).

Therefore, I hope vscode.dev is another step towards making the iPad a viable platform for software development, even if it is restricted to web technologies.

Alongside the release of vscode.dev, Microsoft published a blog post, which describes the new features and framed the target use cases. They used the positioning “Good”, “Better” and “Best”, as outlined below.

Good: For most programming languages, vscode.dev gives you code syntax colourisation, text-based completions, and bracket pair colourisation. Using a Tree-sitter syntax tree, we’re able to provide additional experiences such as Outline/Go to Symbol and Symbol Search for popular languages such as C/C++, C#, Java, PHP, Rust, and Go.
Better: The TypeScript, JavaScript, and Python experiences are all powered by language services that run natively in the browser. With these programming languages, you’ll get the “Good” experience plus rich single file completions, semantic highlighting, syntax errors, and more.
Best: For many “webby” languages, such as JSON, HTML, CSS, and LESS, the coding experience in vscode.dev is nearly identical to the desktop (including Markdown preview).

An important prerequisite for vscode.dev is a modern browser, specifically one that supports the File System Access API.

Unfortunately, as highlighted by the table, mobile browsers are not compatible with the File System Access API, meaning they are limited to individual file uploads/downloads or linking to a remote GitHub or Azure DevOps repository. This is not ideal, but workable in most software development scenarios.

I attempted to use vscode.dev on the iPad to update this blog (JAMstack architecture).

This use case falls between the “Better” the “Best” categories as outlined by Microsoft and I am pleased to report everything worked as designed. Visual Studio Code behaved as expected, delivering a familiar experience to the desktop application. The iPad also worked well, especially when connected to an external keyboard, such as the Apple Magic Keyboard.

However, I encountered two frustrations, specifically the previously mentioned file system access (a constant pain point with Apple iOS and iPadOS), as well as the standard Visual Studio Code keyboard shortcuts, which conflict with the browser keyboard shortcuts.

In theory, both of these challenges could be resolved via a custom browser wrapper (using a lightweight version of Chromium or if Microsoft decided to deliver specific customisations for Microsoft Edge. The use of Microsoft Edge feels like an ideal path for Microsoft, acting as a clear differentiator by exploiting the strengths of their end-to-end ecosystem (e.g. Edge + vscode.dev).

Finally, it should be noted that only a handful of the popular Visual Studio Code extensions are compatible with vscode.dev. Thankfully, this is an area I expect to improve over time.

In conclusion, vscode.dev works well! I can see it being popular within education, where the environment (policy) and/or devices (Chromebooks) have specific limitations.

Regarding the iPad, for web development, the core vscode.dev application works surprisingly well, with a few minor frustrations caused by the restrictions associated with iPadOS. I would be happy to use it today for quick updates/fixes, but the highlighted frustrations would deter me from using it as my daily driver.

With that said, vscode.dev shows great promise and I remain optimistic that the iPad will (one day) become a viable software development platform.

Why Pop!_OS?

Wed, 20 Oct 2021 00:00:00 -0500

Pop!_OS is a Ubuntu-based Linux operating system developed and maintained by System76.

It was first released in 2017 but has quickly established itself as a fast, robust and user-friendly Linux distribution, which is consistently ranked in the top five most popular by DistroWatch. It targets developers, makers, and computer science professionals.

Earlier this year, I transitioned to Pop!_OS, switching from Fedora, which had been my primary Linux distribution since 2011.

I have now been using Pop!_OS for two months on a Dell XPS 17 (9710), supporting my personal and business workloads. This includes eight hour days, covering web (Google Chrome), productivity (LibreOffice, Microsoft Office Web), collaboration (Microsoft Teams, Zoom), development (Docker, GameMaker Studio 2), video/photo editing (OpenShot Video Editor, Krita) and even gaming (Steam, Lutris).

As with any system, there has been the occasional “hiccup”, but I am pleased to report that overall, Pop!_OS has performed admirably, providing a comparable experience to Windows and/or macOS.

With that said, I suspect this experience has been helped by my choice of software, which favours cross-platform and/or free and open-source.

With two months of experience behind me, I thought I would share a few reasons why I believe Pop!_OS is one of the best Linux distributions available today.

Culture

System76 started as a computer manufacturer, specialising in the sale of notebooks, desktops, and servers running Linux (initially Ubuntu). This heritage provides them with a unique perspective on what it takes to make Linux a viable operating system for daily use.

As an example, Carl Richell, founder and CEO of System76 recently joined Jonathan Bennett and Shawn Powers on FLOSS Weekly. The hour-long podcast provided a fascinating glimpse into System76, including their philosophy to design, manufacturing, etc.

In short, the System76 story resonates with me and therefore is something I want to use and promote. Similar to the recently announced Framework Laptop, I feel it is important that ethically and environmentally positive companies that prioritise free and open-source software are allowed to complete, hopefully helping to shape the future of computing.

Installation

A common barrier with Linux is the initial installation, which can be cumbersome for new users, especially if the required open-source software/drivers are not readily available.

A great example is Arch Linux, which is growing in popularity thanks to the rolling release model and the incredible control it provides over the operating system. Although I love the ArchWiki, and Arch User Repository (AUR), the initial installation is certainly not for beginners (prepare to reserve several hours).

Arch-based Linux distributions such as Manjaro can help significantly, but (in my opinion) still fall short of Pop!_OS for speed and simplicity.

System76 provide a dedicated ISO for Intel/AMD or NVIDIA-based systems. This alone is a huge benefit, as installing and maintaining the proprietory NVIDIA drivers on Linux can be a real pain. Once the correct ISO is selected, the installation wizard is fast and intuitive, with options to enable important security controls such as full-disk encryption, etc.

Finally, once at the desktop, Pop!_OS includes a simple getting started guide to configure common settings. In my opinion, this experience is among the best in the business, even superior to Windows and macOS, which have both become bloated over the years with unnecessary steps.

Desktop Envrionment

Pop!_OS ships with a custom version of the GNOME desktop environment called COSMIC. If you do not like GNOME, System76 offer alternatives (e.g. MATE, Cinnamon, KDE Plasma, etc.) However, it should be noted that Pop!_OS was designed with COSMIC in mind.

What I like about Pop!_OS is that COSMIC pre-configures all of the custom settings I would usually have to enable through third-party extensions. For example, I like a very minimal desktop experience, with a vertical layout and a focus on keyboard shortcuts. With Pop!_OS, this can be achieved natively, with almost no effort and zero ongoing maintenance.

Finally, Pop!_OS includes custom power profiles, which are natively integrated into the menu bar. For example, if the system has hybrid graphics (e.g. Intel and NVIDIA), Pop!_OS can be configured in a “single click” to automatically switch depending on the workload.

If you are coming from Windows or macOS, this will sound very trivial, but on Linux, hybrid graphics can be a pain to set up, configure and enable.

Compatibility

Pop!_OS is based on Ubuntu, which itself is based on Debian. I left the Ubuntu ecosystem in 2011 due to the direction of the project (Ubuntu Unity).

However, it is hard to argue against the market strength of Ubuntu and the breadth of Debian-based repositories, delivering the most comprehensive software ecosystem that is often the starting point for new releases.

For example, the recent release of GameMaker Studio 2 for Linux is currently only officially supported on Ubuntu. Although it is possible to make the software work on other distributions, having native support is certainly a time saver.

What I like about System76 is that they have selected the “best bits” of Ubuntu for Pop!_OS, ignoring the more controversial decisions. For example, Pop!_OS does not prioritise the use of Snap packages, instead favouring Debian and Flatpak.

Performance and Reliability

In my experience, Pop!_OS is not the fastest or the most reliable Linux distribution. However, it scores highly in both areas, making it a great balance. For example, I find Arch-based Linux distributions such as Manjaro to be faster, but also less stable.

Pop!_OS follows the Ubuntu twice-yearly update cycle (April and October). As Pop!_OS is not a rolling release distribution (like Arch Linux), it means that software is often a little behind the latest version. For example, Pop!_OS 21.04 shipped with GNOME v3.38.4 (not GNOME 40), however, this conservative update schedule helps ensure any “bleeding edge” bugs or security concerns are (hopefully) resolved before user adoption.

In my experience, Pop!_OS is as stable as Fedora (which is very solid), but also more performant, especially in gaming.

Conclusion

In conclusion, Pop!_OS just works!

It allows me to focus on my workloads (e.g. productivity, collaboration, development), instead of the operating system itself. In my opinion, this is the true test of a great operating system, which should really “fade into the background”, acting as the vehicle to “get stuff done”.

I would happily recommend Pop!_OS to any Linux user, either someone starting new or an experienced user simply looking for a robust daily driver. In either scenario, I think you would be hard-pressed to find a more well rounded Linux distribution.

DevOps @ Enterprise Forum

Fri, 08 Oct 2021 00:00:00 -0500

Earlier today, I presented at the DevOps @ Enterprise Forum hosted from Poland.

My session was focused on Infrastructure-as-Code, targetting IT managers (not architects/engineers) looking to learn about the concepts, prerequisites, implementation approach, and the associated value proposition.

I shared my experience covering the Methodology and Foundational IT Ecosystem, as well as DevOps Culture, DevOps Operating Model, Immutable Infrastructure, and of course, Infrastructure-as-Code itself.

The session was virtual and predominantly hosted in Polish (I do not speak Polish), therefore community connections were limited. However, I have had great engagement post-event and it has been great fun collaborating with passionate individuals.

If you are interested in DevOps Culture or concepts/technologies such as Immutable Infrastructure and Infrastructure-as-Code, feel free to connect using one of the usual channels (Email, Twitter, LinkedIn, etc.)

The recording for the session should be available online soon. I will update this post once it has been published.

Thanks to the team at DevOps @ Enterprise Forum, as well as everyone that attended!

Windows 11 VM

Tue, 05 Oct 2021 00:00:00 -0500

Earlier today, Microsoft officially released Windows 11. As highlighted in my previous article, Windows 11 does not focus on technical innovation and instead should be considered a “marketing release”, intended to re-energise the faltering partner market (e.g. Dell, HP, etc.)

With that said, Windows 11 does include some interesting incremental changes that go beyond Windows 10 21H1, most notably the new simplified Start Menu.

It remains unclear what new features (if any) will be delivered to Windows 10 in the future, therefore Windows 11 is the logical path forward for most users.

Windows 11 can be installed from an ISO or upgraded via Windows Update. However, in either scenario, Windows will complete a compatibility check to ensure the host hardware is compatible.

Outside of the arbitrary processor requirements (Intel 8th Generation / AMD Ryzen 3000 Series), the most controversial requirement is the Trusted Platform Module (TPM) v2.0.

TPM v2.0 is an international standard for a secure cryptoprocessor, which was first introduced in 2014, therefore any system built after this point should be compatible. However, the reality for budget laptops, desktops and/or custom-built systems is more convoluted.

Interestingly, the TPM v2.0 requirement is not a technical limitation of Windows 11, instead, a requirement enforced by Microsoft through code. Arguably, this position demonstrates Microsoft’s dedication to security, however, it is difficult to ignore that it also acts as a convenient cut-off point to force hardware upgrades.

An example of where this requirement can be a challenge is client virtualisation. Although it is possible to virtualise TPM v2.0 for compatibility, this feature is not supported by all hypervisors. Therefore, if you need to run Windows using client virtualisation, you might encounter issues.

Thankfully, as TPM v2.0 is not a technical limitation of Windows 11, the bypass mechanism is relatively simple.

I previously shared a simple script created by “AveYo”, which acts as a wrapper for the official Microsoft Media Creation Tool, allowing the user to select any version of Windows 10 (1507 to latest) or 11 (21H2 to latest). By default, the Microsoft Media Creation Tool will always select the latest version, which may not be the desired outcome.

If Windows 11 is selected, the wrapper automatically disables the TPM v2.0 requirement, allowing the resulting ISO to be installed on a wide variety of hosts (hardware or virtualised).

I recently used the wrapper to create a Windows 11 virtual machine for my Linux (Pop!_OS) host, running on VMware Workstation Player.

Outside of the TPM v2.0 requirement, everything else works as originally designed, including Windows Update, etc.

Deloitte Experience Analytics

Fri, 01 Oct 2021 00:00:00 -0500

Earlier today, Deloitte published the on-demand videos from the Experience Analytics event.

I presented with the team from Boomi as part of the session “Achieving Continuity with Modernisation to Create a New and Improved IT Ecosystem”.

Within the session, I shared our customer journey and a few examples of how integration technologies played a critical role in enabling a series of highly complex acquisition and divestiture activities. Outlined below is a short extract from the session, which provides additional framing.

“In September 2019, we announced our plan to divest from our parent company. What made this unique was the requirement to re-build IT from the ground up, covering the architecture, technology, processes and the organisation.

As someone that has spent his career managing legacy, technical debt, and inertia, the ability to rebuild everything, covering workplace services, hosting, networking, as well our applications and data services was an exciting prospect.

We had a two-year window to execute the change, whilst maintaining business continuity throughout.

To make things more interesting, during the divestiture, we demonstrated our ambition as a standalone company, with a major acquisition. From an IT standpoint, this essentially doubled the size of the IT ecosystem.

This was obviously all completed whilst battling against a global pandemic, where travel was restricted, and supply chains were heavily impacted.”

To access the on-demand videos, you must first register for the event. Once complete, all of the content is made available, including access to the community of speakers and participants.

Overall, it was great to re-engage with the “tech” community and I look forward to the next opportunity.

Boomi

Wed, 29 Sep 2021 00:00:00 -0500

Following the Deloitte Experience Analytics event last week, the team at Boomi have released the associated customer story.

As discussed at the event, the Customer Story describes how we balanced business continuity with modernisation to create a new and improved IT ecosystem.

I highly recommend watching the short video infographic (link below), which highlights the key milestones and value proposition.

Boomi have been a great partner, helping us to unlock agility and speed to value via their cloud-native Integration Platform-as-a-Service (iPaaS).

CTO Confessions

Tue, 28 Sep 2021 00:00:00 -0500

UPDATE - The CTO Confessions episode “Proper Animal Care Through Tech Innovation with Matt Bull” is now available to download/stream.

I am pleased to share that I will be a guest speaker on an upcoming episode of the CTO Confessions podcast, hosted by IT Labs.

IT Labs is a US-based software company that helps businesses innovate across Cloud, DevOps, and AI/ML.

Promoting the company culture, IT Labs are also very active in the community. For example, the CTO Confessions podcast launched in April 2020, but already has over 60 episodes, including guest speakers from Google, Tui, Accenture, easyJet, CloudBees, Revolut, etc.

The podcast, alongside the other community content, is available for free. Therefore, I highly recommend subscribing via one of their channels.

As with previous episodes, I expect the discussion to focus on career and IT leadership, as well as architecture/technology trends.

I am looking forward to collaborating with IT Labs, with the episode scheduled to be published in December.

Experience Analytics

Sun, 12 Sep 2021 00:00:00 -0500

Update - The Deloitte Experience Analytics session is now available to view online.

Later this month I will be presenting at the Deloitte Experience Analytics event in London, which is focused on AI and analytics in business.

The event synopsis can be found below.

Experience Analytics is Deloitte’s flagship conference for data, analytics, and artificial intelligence in business. This year, we’re hosting six live sessions across 20-22 September, delivered and attended by C-Suite and Senior Executives to discuss the latest insights and trends in AI and analytics.

Through a series of Ted-X style talks and live demonstrations, we’ll explore the key issues, challenges and opportunities presented by AI and data. We’ll also focus on how businesses can utilise the latest AI technologies to become an insight-driven organisation.

While we were restricted to virtual live streams in 2020, we’re delighted to be able to welcome a small number of you to a physical venue this year. For those unable to attend, we will be live streaming each of our sessions so that no one will miss out!

I will be speaking on day three with the team from Boomi as part of the session “Achieving Continuity with Modernisation to Create a New and Improved IT Ecosystem”.

I plan to share details of our journey over the past eighteen months, which was highlighted in the article “Modern IT Ecosystem”.

The event runs from 20 September to 22 September and includes virtual participation.

Linux Distributions

Sun, 05 Sep 2021 00:00:00 -0500

Over the past two decades, I have hopped between different Linux distributions, eventually landing on Fedora in 2011.

Earlier this year, I decided to take a look at some other Linux distributions. This evaluation was driven by personal intrigue and not any specific issue with Fedora.

In this article, I wanted to share a summary of my findings, which has led me to what I believe to be the “best” three Linux distributions. It should be noted, this conclusion is very subjective and my choices are “GNOME-centric”, which is a personal preference that I respect would not suit everyone.

Pop!_OS

Pop!_OS, developed by System76, is based on Ubuntu and features a custom GNOME desktop, known as COSMIC (Computer Operating System Main Interface Components).

In my opinion, Pop!_OS achieves the best “out of the box” experience for Linux, covering Intel, AMD and NVIDIA hardware. It delivers the closest equivalent to the Apple “it just works” mantra, making it an ideal choice for anyone that is looking to get up and running with minimal effort.

Knowing that Pop!_OS is based on Ubuntu (which itself is based on Debian), there is a huge developer/support community, with thousands of compatible software packages (over 68,000) and knowledge articles covering almost every conceivable scenario.

Pop!_OS includes two major releases per year (mimicking Ubuntu), with incremental updates delivering new features, patches, etc. This is a “tried and tested” update methodology, which delivers a highly consistent and reliable experience.

In addition to the standard Ubuntu features, Pop!_OS includes some unique capabilities, such as a custom installer, full disk encryption, power profiles, firmware management, and integrated hybrid graphics switching. These capabilities help to make Pop!_OS feel like an operating system that could rival Windows and macOS as a consumer product targeting the general public.

If you are looking for a “no-fuss” Linux distribution that is well optimised for productivity, gaming and development workflows, I would highly recommend Pop!_OS.

Manjaro

Manjaro is based on Arch Linux, which is a “rolling-release” distribution, meaning there are no “major” releases (like with Pop!_OS), with updates being delivered continuously (rolling) throughout the life of the product.

Although Majaro is based on Arch Linux, it dramatically simplifies the installation and support process, providing an opinionated software stack. For example, Manjaro includes different editions, specifically Xfce, KDE Plasma or GNOME.

Updates tend to be made available at a slower pace than Arch Linux, but are in theory more heavily tested to ensure interoperability and reliability.

Similar to Pop!_OS, Manjaro has great support for open and proprietary drivers, meaning NVIDIA graphics can be easily installed and maintained. Manjaro also has access to the Arch package repositories and Arch User Repository (AUR) which contain thousands of software packages, comparable to Debian. Package management on Arch-based Linux distributions is very different to Debian-based distributions but arguably more powerful and versatile. Manjaro includes a GUI application known as Paman which simplifies package management, covering all common use cases.

Another benefit of Manjaro is the ArchWiki, which is the ultimate Linux knowledge base. The level of detail, breadth of content and ease of consumption is unrivalled.

If you are looking for a rolling release Linux distribution, without the complexity of Arch Linux, Manjaro is a great choice.

Fedora

Fedora, sponsored by Red Hat, is an upstream source of Red Hat Enterprise Linux (RHEL). As a result, it is an incredibly robust Linux distribution, with a core that is “battle-tested” for enterprise use.

Major Fedora releases occur every six months, which tend to include the “latest and greatest” updates, covering the Linux Kernal and software dependencies. Fedora is passionate about open-source software, therefore does not come pre-installed with any proprietary software. This is a great philosophy but can be a pain when looking to use hardware that does not include open-source drivers (e.g. NVIDIA graphics, etc.)

As a result, Fedora delivers a very “pure” Linux experience, with minimal shell customisations (unlike COSMIC with Pop!_OS).

If you are looking for a rock-solid Linux distribution that focuses on cutting edge features and open-source software, Fedora is a great choice.

Conclusion

That’s it! Depending on your requirements, I believe these three Linux distributions represent the very best available. If I had to select one, I would go with Pop!_OS, which delivered a great balance of performance, reliability and compatbility.

GMS2 Xbox Development

Thu, 02 Sep 2021 00:00:00 -0500

Since March, I have been building a platformer game called Ninja Caves for my son using GameMaker Studio 2. I have documented my progress across a series of articles.

The game is currently available for Windows, macOS, and iPadOS. However, following the recent GameMaker Studio 2 license changes, I had the opportunity to export the game for the Universal Windows Platform (UWP).

The Universal Windows Platform targets a common API on all devices that run Windows 10 and can be distributed via a unified app store, covering PC, Xbox, HoloLens, Surface Hub, etc. Therefore, in theory, I should be able to export my game as a Universal Windows Platform app and have it run natively on an Xbox.

To get started, the following four prerequisites must be met:

GameMaker Studio 2 running on Windows 10 (Free)
GameMaker Studio 2 Indie License (£7.19)
Visual Studio Community Edition (Free)
Xbox Creators Program License (£14.40)

The GameMaker Studio 2 Indie License grants access to export to the Universal Windows Platform. The license is paid monthly but is only required at the point of export (development is free).

The Xbox Creators Program allows developers to self-publish games for Xbox (including Xbox Series S and Xbox Series X) via a simplified certification process. This is a great path for independent developers, learning, education, etc.

The program also supports the use of a consumer Xbox to be used for development via the “Xbox Developer Mode”, removing the need to acquire a dedicated Xbox development kit.

Once the prerequisites have been met, GameMaker Studio 2 can be configured to compile the game source code as a Universal Windows Platform app.

This process requires Visual Studio (not Visual Studio Code), with the “Universal Windows Platform Development” workload and the “v142 tools” component to be installed.

The linked guide provides an excellent overview of the setup process but does not include the locations for Visual Studio Community 2019.

If you are using Visual Studio Community 2019 on Windows 10, open GameMaker Studio 2 and select “File > Preferences > Platform Settings > Windows UWP”. Set the following locations:

Visual Studio Location: C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\Common7\Tools\VsDevCmd.bat
Win UWP Tools Location: C:\Program Files (x86)\Windows Kits\10\bin\10.0.19041.0\x86

NOTE: These locations might be different depending on your version of Windows and Visual Studio.

With these locations configured, the rest of the guide should work without issue, covering the enablement of “Windows 10 Developer Mode”, as well as the setup and registration of the “Device Portal for Desktop”.

You should now be able to run/export your game from GameMaker Studio 2 as a Universal Windows Platform app.

If you want to test on Xbox hardware, you will need to acquire a consumer console and enable “Developer Mode”. This is a simple process, which does not void your warranty and can be switched on-demand (between developer and retail).

With developer mode enabled, any compatible Universal Windows Platform app can be transferred to the console and executed. This can be achieved via a network transfer using the Developer Portal.

The photo below highlights Ninja Caves running as a Universal Windows Platform app on an Xbox One Digital Edition. I have “Picture-in-Picture” mode enabled on my monitor, therefore the left side is showing my Windows 10 development environment, with the right side showing ninja Caves running on the Xbox One.

Ninja Caves runs well on the Xbox One, although I did notice a few minor input bugs and sub-optimal performance. Recognising that I did not optimise the code for this platform, I am impressed with the results of the default export.

As I continue the development of Ninja Caves, I may look to release the game on the Windows Store as a Universal Windows Platform app.

Manjaro Full Disk Encryption

Tue, 24 Aug 2021 00:00:00 -0500

Manjaro is a great Linux distribution, delivering an opinionated Arch Linux experience that simplifies the setup and daily management.

One frustration, unlike Debian-based Linux distributions (e.g. Ubuntu, Pop!_OS), there is no “one-click” option to enable full disk encryption as part of the initial setup.

In 2021, I consider full disk encryption a foundational requirement, especially with modern hardware-accelerated encryption that mitigates any historical performance impact.

Thankfully, there is a workaround for Manjaro. It is not perfect, but certainly better than no encryption.

The approach is to manually partition the disk to include a small unencrypted boot partition. This partition is only used to boot the operating system, with all other partitions (storing the operating system, data and swap) being fully encrypted.

As part of the Manjaro setup process, when using the default installer, select “Manual Partitioning” from “Partitions”.

Delete all existing partitions and configure the following.

Partition 1

Size: Minimum 300MB (Recommened 2GB)
File System: fat32
Mount Point: /boot/efi
Flags: boot

Partition 2

Size: System RAM + 5%
File System: linuxswap + Encryption (luks)
Mount Point:
Flags: swap

Partition 3

Size: Remaining Available Space
File System: ext4 + Encryption (luks)
Mount Point: /
Flags: root

The SWAP partition is optional but recommended. If you plan to use hibernation, ensure you allocate enough capacity to cover your system memory, including a 5% buffer. For example, my system has 64GB RAM, therefore I allocate 70GB for SWAP.

The Linux Unified Key Setup (LUKS) is a platform-independent disk encryption specification, which is why any encrypted partition will show as “LUKS” instead of the chosen filesystem (e.g. EXT4).

LUKS is a fully open-source tool that has been the standard for disk encryption in Linux environments for many years. It sits in the kernel layer and encrypts storage at a disk block level, allowing for the transparent deployment of any file system.

Once configured, Manjaro will install as normal. Following a cold boot, the GRand Unified Bootloader (GRUB) will prompt for the disk encryption password.

Assuming the password is accepted the encrypted partitions will be opened and the Linux kernel will load as normal.

There are three minor challenges with this approach.

As the encryption password is requested by GRUB, before the operating system, any user-defined keyboard preferences are not loaded. Therefore, if your password includes special characters, you must know their location on a default US QWERTY keyboard.
Cold boot performance is slow, taking up to 30 seconds to accept the encryption password and open the encrypted partitions. Thankfully, this is only the case with a cold boot, which feels like a small price for encryption.
GRUB is not sophisticated when requesting the password. If the password is entered incorrectly, you must restart and try again.

Hopefully, the Majaro team will improve the full disk encryption experience over time, delivering a one-click option as part of the initial setup that also covers the boot partition.

Pop!_OS Setup

Fri, 20 Aug 2021 00:00:00 -0500

I recently switch to Pop!_OS as my primary Linux distribution. So far, I have been impressed with the speed, reliability and user experience.

In 2019, I shared my setup process for Linux, at the time, Fedora 31. This article will follow a similar approach, focused on my setup process for Pop!_OS 21.04.

Once Pop!_OS is installed, I immediately run the following three terminal commands to confirm my Linux kernel version, as well as update and upgrade the software.

uname -r

sudo apt-get update

sudo apt-get upgrade

To help track ongoing system information I install Neofetch using the following terminal commands. Neofetch displays information about the operating system, software and hardware. For example, you can track the number of installed software packages (Debian and Flatpak) and memory utilisation.

sudo add-apt-repository ppa:dawidd0811/neofetch

sudo apt-get update && sudo apt-get install neofetch

If an NVIDIA graphics card is installed, I use the following command to confirm the latest proprietary driver is installed.

nvidia-smi

If required, update the driver from the “Pop!_Shop” by clicking the “Installed” tab. If a new driver is available, it should be shown in the list.

When using a laptop with NVIDIA Optimus technology, I switch the operating system to “Hybrid Graphics” by selecting the “Battery” dropdown in the Top Bar. Enabling this setting will require a system restart, but should help balance battery life and performance.

Once complete, I make a few changes to the desktop to suit my workflow. Thankfully, the Pop!_OS defaults are already well optimised, therefore I do not install any third-party extensions, etc.

In “Settings”, I configure the following options. Any option not specified is left default:

Desktop > General > Super Key Action = Workspaces
Desktop > General > Top Bar > Show Workspace Button = Off
Desktop > General > Top Bar > Show Applications Button = Off
Desktop > Appearence = Dark
Desktop > Dock > Extend Dock = Off
Desktop > Dock > Show Launcher = Off
Desktop > Dock > Show Workspaces = Off
Desktop > Dock > Dock Visibility > Alway Hide = On
Desktop > Docker > Dock Size > Customer Size = 24
Desktop > Position on the Desktop = Left
Desktop > Workspaces = Fixed Number
Desktop > Workspaces > Number of Workspaces = 2
Desktop > Workspaces > Placement of Workspace Picker = Right

With the Pop!_OS desktop configured to suit my workflow, I start installing software. The following steps include the installation process for my common software.

Open a browser (e.g. Firefox) and download the following Debian (.deb) packages. Alternatively, add the relevant repositories via the terminal.

Assuming the installers were downloaded, open the terminal and navigate to the “Downloads” folder. Install each application individually using the following command.

sudo apt-get install ./<Installer Filename>

Once complete, open the “Pop!_Shop” and install the following applications. Where available, I select the native “Pop!_OS (deb)” package. Flatpak is a great package manager, but not always as tightly optimised for the specific Linux distribution.

I now install any application that is not available from within the “Pop!_Shop”, starting with Docker. Open the terminal and run the following commands.

sudo apt-get update

sudo apt-get install \
    apt-transport-https \
    ca-certificates \
    curl \
    gnupg \
    lsb-release

curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg

echo \
"deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu \
$(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

sudo apt-get update
sudo apt-get install docker-ce docker-ce-cli containerd.io

Alongside Docker, I also install Docker Compose.

sudo curl -L "https://github.com/docker/compose/releases/download/1.29.2/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose

sudo chmod +x /usr/local/bin/docker-compose

Once complete, I install VMware Workstation Player, which is my preference when running local virtual machines (e.g. Windows, etc.)

sudo apt-get update

sudo apt install build-essential

wget --user-agent="Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Firefox/60.0" https://www.vmware.com/go/getplayer-linux

chmod +x getplayer-linux

sudo ./getplayer-linux --required --eulas-agreed

Finally, I run the following commands to enable the embedded firewall and install/enable restricted formats (DVD, MP3, QuickTime) as well as Microsoft Core and ClearType fonts.

sudo ufw enable

sudo apt-get update && sudo apt-get install ubuntu-restricted-extras

sudo add-apt-repository multiverse

sudo apt update && sudo apt install ttf-mscorefonts-installer

wget https://gist.github.com/maxwelleite/10774746/raw/ttf-vista-fonts-installer.sh -q -O - | sudo bash

wget https://gist.githubusercontent.com/maxwelleite/913b6775e4e408daa904566eb375b090/raw/cbfd8eb70184fa509fcab37dad7905676c93d587/ttf-ms-tahoma-installer.sh -q -O - | sudo bash

sudo fc-cache -f -v

With all my software installed, I start each application and complete any required setup process.

I usually start with 1Password, followed by any application that requires data to be downloaded (e.g. Dropbox). The majority of applications will automatically detect the operating system theme, if not, manually select “dark” mode from the application preferences/settings/options.

To enhance Ulauncher, I add the “File Search” and “System” extensions. This can be done from “Ulauncher Settings”, specifically the “Extensions” tab.

To conclude, I download the latest Windows ISO and create a local Windows virtual machine using VMware Workstation Player. The only application I install in the virtual machine is Microsoft Office.

With these steps complete, I run one final update and upgrade.

sudo apt-get update

sudo apt-get upgrade

Pop!_OS is now ready for action! Enjoy!!!

Proton - The Trojan Horse

Sat, 07 Aug 2021 00:00:00 -0500

Last month, Valve announced the Steam Deck, a new handheld gaming computer that is set to be released in December 2021.

The hardware itself has generated a lot of interest from the community, driven by the modern AMD Zen 2 / RDNA 2 architecture and the “open” nature of the platform, which is essentially a PC with an integrated controller.

I do not plan to cover the hardware specification in this article, which has been thoroughly covered by the usual suspects (IGN, Digital Foundry, etc.)

With that said, I would recommend watching the Linus Tech Tips video (embedded below), which is a fun “rapid-fire” hands-on of the hardware and software.

Although I am always excited about the release of new hardware, it is the software direction that has me most intrigued.

It is well known that Valve has a love/hate relationship with Microsoft. In 2012, Gabe Newell (Valve CEO) publically expresses his frustration and concern regarding the direction Microsoft was taking the Windows ecosystem, essentially creating a “walled garden”.

As a result, Valve has been slowly increasing its commitment to Linux, with the release of Steam Client for Linux, SteamOS (Linux-based), Steam Machines, etc. They have also contributed to several open-source projects focused on Linux software compatibility.

Although not restricted to a specific operating system, the Steam Deck will ship with a new version of SteamOS, which is a modified version of Arch Linux.

To ensure compatibility, Valve has made some bold claims regarding Proton, which is an open-source project developed by Valve and CodeWeavers that provides a Linux compatibility layer for Windows games.

Proton is a fork of the popular Wine compatibility layer, which was first introduced in 1993. The name Wine stands for “Wine Is Not an Emulator”, which is an important distinction, as it aims to translate (not emulate) Windows API calls into native POSIX calls. This approach eliminates the performance and memory penalties of methods such as emulation and virtualisation.

Proton focuses on graphics APIs, achieved via Direct3D-to-Vulkan translation layers, specifically DXVK for Direct3D 9, 10, 11 and VKD3D for Direct3D 12. The goal is to deliver compatibility and (where possible) native performance.

The video embedded below from Tom (GloriousEggroll) provides a great overview of how proton works, including the key components and game examples.

In 2018, Valve integrated Proton into the Steam client, known as “Steam Play”. The results are impressive, but not perfect. For example, many games run poorly or fail to run at all. The website ProtonDB tracks compatibility across 19,000+ games, with 15,000+ reported as working.

However, as part of the Steam Deck announcement, Valve has stated its intent to achieve 100% compatibility of the entire Steam Library ahead of the Steam Deck launch in December 2021.

This is a highly ambitious outcome, but not without precedent, knowing that Valve has been actively exploring this space since 2012.

Assuming success (or 90%+), Proton could help re-position Linux as a first-class operating system for gaming, with the Steam Deck acting as a trojan horse to showcase the possibilities.

The ramifications could be profound, further reducing the reliance on Microsoft Windows, which would be a win for hardware vendors and consumers, recognising the free and open-source nature of Linux, as well as the inherent privacy and security benefits.

In conclusion, I am eager to see the Steam Deck succeed, not because of the great hardware, but what it could mean for the software ecosystem. considering the history, it is also safe to assume that Valve has a similar ambition, providing unrestricted access to their Steam digital distribution service.

GameMaker Studio 2 Linux

Fri, 06 Aug 2021 00:00:00 -0500

As a hobbyist independent game developer, I am a long-time user of the Game Creation System (GCS) GameMaker Studio 2 by YoYo Games (owned by Opera).

GameMaker Studio 2 targets 2D games, which means it is not as versatile as other popular game engines such as Unreal Engine, Unity, or Godot.

However, it is very beginner-friendly and includes excellent multi-platform publishing support, making it a great choice for anyone looking to create platformer or top-down games, such as shooters, role-playing, racers, Metroidvania, bullet hell, etc. It is also great for prototyping, thanks to the highly accessible, integrated workflow.

As an example, I am currently working on a 2D platformer called Ninja Caves, developed for my son (six years old). The game is inspired by Crystal Caves and is available for Windows, macOS and iPadOS.

I have documented the development process in the following articles.

Although GameMaker Studio 2 can publish games for Linux, the IDE has historically been restricted to Windows and macOS.

Earlier today, YoYo Games announced the release of GameMaker Studio 2 (Beta) for Ubuntu.

I was personally excited by this announcement, as it takes me one step closer to my platform-agnostic (e.g. Windows, macOS, Linux) goal for my common software library.

At this time, YoYo Games are only offering a Debian Linux package, with support for Ubuntu. However, as I recently switched to Pop!_OS as my primary Linux distribution, which is Ubuntu-based, the package should be compatible.

I am pleased to report the installation was completed successfully, with no errors or warnings. Although I did need to download the package from the GameMaker Studio Community forum via Firefox, as it refused to download using Chrome.

With the latest runtime installed (v23.1.1.329), I loaded my active project, Ninja Caves. The source code for Ninja Caves can be found on GitHub.

I would position Ninja Caves as a “medium complexity” project, targeting multiple platforms (Windows, macOS, Linux, iPadOS, tvOS), with two primary input mechanisms (keyboard/mouse and controller).

Ninja Caves includes a wide range of custom assets, specifically Fonts, Objects, Rooms, Scripts, Shaders, Sounds, Sprites, and Tile Sets. All assets are present and correct on the Linux build, with no obvious bugs or limitations.

Ninja Caves also incorporates particles and GPU controller functions, specifically extended blend modes to enable “baked lighting” across the levels. These features can be temperamental when working across multiple platforms, but I am pleased to report that they appear to be working as designed within the IDE and the VM (interpreted code).

The only issue I have identified (so far) relates to the embedded debugger, which currently fails to run. The debugging overlay itself launches, but the game runner fails to load. If I build the game outside of the debugger, everything works as designed.

Finally, although I have not objectively tested the performance of the IDE or build process, it “feels” responsive. Similar to what I experience on Windows and macOS. This is a great outcome, considering the pre-release nature of the build.

In conclusion, the release of GameMaker Studio 2 for Linux, alongside the recent announcement from Valve regarding their Linux-powered Steam Deck, makes me excited about the future of gaming (and software in general) on Linux!

I am not yet ready to declare this “the year of the Linux desktop”, but things are certainly trending in the right direction.

UPDATE (10-AUG-2021)

Earlier today, YoYo Games announced an update to the GameMaker Studio 2 license model.

In summary, they have removed the existing perpetual licensing options, replacing them with a three-tier subscription model (monthly or annual).

Free (£0 pm) - Develop games with no restrictions, but cannot publish/export.
Indie (£7.19 pm) - Develop games with no restrictions, publish/export to Desktop, Mobile, Web, and UWP.
Enterprise (£62.99 pm) - Develop games with no restrictions, publish/export to Console, Desktop, Web, Mobile, Web, and UWP.

Ordinarily, I would be nervous about any switch to subscription-based licensing. However, knowing that the base tier is free and unrestricted for development, provides a lot of flexibility. For example, a license is only required to publish/export a game, which for many hobbyist/independent developers could be covered by a single month.

As with any change of this type, I am sure YoYo Games will receive a lot of community feedback, therefore, I would recommend keeping an eye on the following channels.

Finally, it is worth noting that anyone with a perpetual license (like me) will retain that license for the life of GameMaker Studio 2, even if you decide to “upgrade” or purchase a new subscription license.

YoYo Games are also offering incentives to licence holders, allowing for multi-month trials of the new subscriptions tiers. For example, I have a free six-month trial for Enterprise, which unlocks the ability to publish/export for consoles.

Ninja Caves - Update Two

Thu, 05 Aug 2021 00:00:00 -0500

Update - Ninja Caves is no longer maintained. As a result, the Apple App Store (iPadOS) version has been retired. The Windows and macOS versions are still available for download, but they may not work correctly with modern hardware/software.

Since March, I have been building a platformer game for my son (six years old), inspired by Crystal Caves, using GameMaker Studio 2.

I have been documenting my progress across a series of articles.

Today I released version 1.0.7 across for Windows, macOS and iPadOS.

The release includes new levels, features and bug fixes. The video below highlights a recent playtest, providing an insight into the first seven levels, including some of the new features.

I have outlined some of the new features below, which continue to be inspired by legendary platformer games from the past.

Abilities: New abilities, which can be collected on specific levels and last for the duration of the level. For example, super jump and glide, which provide new methods to navigate the terrain.
Equipment: The player can now equip a torch instead of the Ninja Stars, which can be used to light dark areas. The torch is connected to the right analogue stick and/or mouse, allowing the light to be directed.
Explosives: New obstacles (rocks), which cannot be destroyed using the Ninja Stars. However, dynamite can be used to destroy the objects via a TNT blasting machine (like from Looney Toons).
Weather: New weather effects, including fog and wind. The fog appears close to the ground, with subtle movement, while the wind blows horizontally across the entire level. The weather effects do not (currently) have an impact on the gameplay, but add to the atmosphere of the game.

The weather effects use the game engines embedded particle system, which enabled a surprisingly simple implementation, requiring only two custom events.

The code below was taken from the “create” event, defining the parameters of the required particle effect.

time = 30;
alarm[0] = time;

particle_system = part_system_create();

particle_fog = part_type_create();

part_type_shape(particle_fog, pt_shape_cloud);
part_type_size(particle_fog, 1, 1, 0, 0);
part_type_direction(particle_fog, 180, 180, 0, 0);
part_type_speed(particle_fog, 0.01, 0.02, 0, 0);
part_type_alpha3(particle_fog, 0.03, 0.1, 0.03);
part_type_life(particle_fog, 240, 240);

The code below was taken from the “alarm” event, which triggers the particle effect.

alarm_set(0, time);

var camera_x = camera_get_view_x(view_camera[0]);
var camera_y = camera_get_view_y(view_camera[0]);

var camera_width = camera_get_view_width(view_camera[0]);
var camera_height = camera_get_view_height(view_camera[0]);

repeat(20) {
    var random_y = irandom_range(0, 200);
    var random_x = irandom_range(0, camera_width);

    part_particles_create(particle_system, camera_x + random_x, camera_y + camera_height - random_y, particle_fog, 1);
}

Based on my current implementation, the in-game effect is subtle but provides the foundation for more impressive particle effects in the future.

The photo below is of my son and daughter playtesting the new levels and features.

In an upcoming release, I plan to overhaul the menu system, recognising my current implementation is very rudimentary.

Infrastructure-as-Code

Wed, 21 Jul 2021 00:00:00 -0500

UPDATE - Please refer to the article “DevOps @ Enterprise Forum” for additional details.

On the 8th October, I will be presenting at the DevOps @ Enterprise Forum in Poland.

The conference agenda includes examples of industries and companies that have successfully implemented DevOps, as well as sessions to collaberate with peers.

I have been asked to present on the topic of Infrastructure-as-Code.

The synopsis of my session can be found below.

Over the past decade, there has been a heavy focus on the cloud. However, a cloud strategy alone is not enough to unlock business value… Automation plays a critical role, helping to streamline development and operations, improve the user experience, as well as guarantee security, quality, and privacy.

In this session I will share insights from our DevOps / NoOps journey, with a focus on our automation strategy, specifically Infrastructure as Code.

It has been over two years since I last presented at a conference, therefore I am looking forward to reconnecting with the community.

APT Cheat Sheet

Mon, 12 Jul 2021 00:00:00 -0500

Advanced Package Tool (APT) is a package manager for Unix-like computer systems that facilitates the retrieval, configuration and installation of software packages. APT is used by Debian-based Linux distributions, such as Ubuntu, Pop!_OS, Linux Mint, etc.

dpkg (Debian Package) is a low-level tool, at the base of the package management system. APT is a higher-level tool, which is more commonly used than dpkg as it can fetch packages from remote locations and automatically manage package dependencies, etc.

I am most familiar with Dandified Yum (DNF), which is an alternative package manager for RPM packages. This is due to my history with Fedora, which uses DNF/RPM by default.

However, in recent weeks, I have switched to Pop!_OS, which is Ubuntu-based and therefore uses APT.

Similar to my “DNF Cheat Sheet” article, the goal of this article is to highlight common APT commands (acting as a cheat sheet).

Before proceeding, it is important to understand the difference between a repository and a package.

A repository containers packages.
A package contains an application.

Repositories and packages simplify software distribution and installation.

List All Repositories

The following command will list all repositories checked by APT.

apt-cache policy

List All Packages

The following command will list all packages from installed repositories.

apt-cache pkgnames

List Installed Packages

The following command will list all installed packages.

apt list --installed

Install Package

The following command will install a specific package.

sudo apt-get install <package-name>

Remove Package

The following command will remove a specific package.

sudo apt-get remove <package-name>

Remove Package and Configuration

The following command will remove a specific package and the related configuration.

sudo apt-get --purge remove <package-name>

Remove Orphen Packages

The following command will remove any packages that are no longer required.

sudo apt-get --purge autoremove

Check for Updates

The following command will update the list of available packages and their versions, but it does not install or upgrade any packages. The “update” command should be run before the “upgrade” command.

sudo apt-get update

Upgrade All Packages

The following command will upgrade installed packages. The “update” command should be run before the “upgrade” command.

sudo apt-get upgrade

View Package Information

The following command will show the details of a specific package.

apt show <package>

Additional information regarding APT can be found on the Ubuntu documentation.

Ninja Caves - Release

Tue, 06 Jul 2021 00:00:00 -0500

Since March, I have been building a platformer game for my son (six years old), inspired by Crystal Caves, using GameMaker Studio 2.

I have been documenting my progress across a series of articles.

Today we hit an important milestone… The Minimum Viable Product (MVP) release!

The initial release is available for Windows and macOS and can be downloaded for free from LifeinTECH (links below). In the future, I plan to make the game available via Steam.

The code has not been signed (yet) by Microsoft or Apple, therefore users might receive a security warning at the first launch.

I have also submitted Ninja Caves for release on Apple iPadOS (Apple tvOS planned in the future), but am currently waiting on the review process. The image below shows the Apple App Store Connect process.

The initial release includes the first four levels and is playable on Windows and macOS using the keyboard/mouse or controller. The game requires a controller on iPadOS (no virtual controller support, yet).

The video below showcases the first four levels, with developer commentary highlighting some of the features and design decisions.

As previously stated, Ninja Caves was developed with (and for) my six-year-old son, therefore the difficulty has been balanced based on his skill set. Therefore, I have pitched the game for players new to the platform genre.

Additional information regarding Ninja Caves, including the Privacy Policy, can be found linked below.

Ninja Caves - Privacy Policy

Feel free to download the game, all feedback welcome. I hope you enjoy it!

UPDATE (13-JUL-2021)

Following a prolonged review process with Apple, I am pleased to report that Ninja Caves is now available for iPadOS on the iOS App Store for free.

It is important to note that Ninja Caves requires an external (Bluetooth) controller. The game does not currently include a virtual controller.

Over the coming months, I plan to add new features and levels. All will be made available for free (no ads or in-app purchases) on Windows, macOS, iPadOS and tvOS (coming soon).

My Setup (Q3 2021)

Thu, 01 Jul 2021 00:00:00 -0500

Update - Updated to include the 2021 MacBook Pro, released in October 2021.

Final Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

Home Setup

At home, I switch between three devices (one desktop PC and two laptops).

The full specification can be found below:

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
NVIDIA GeForce RTX 3090 FE 24GB GDDR6X
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

The full specification can be found below:

16-inch MacBook Pro (OCT-2021)
Apple M1 Max (10-core - 8 Performance / 2 Efficiency)
32-core GPU (10.4 Teraflops)
16-core Neural Engine
32GB Unified Memory (400GB/s Memory Bandwidth)
1TB SSD (7.4GB/s Read)
3x Thunderbolt 4, HDMI, 3.5mm Headphone, SDXC Card Reader, MagSafe 3
16.2-inch Liquid Retina XDR Display (3456x2234 @ 120Hz, 1600nits)

I also have a secondary laptop, the Dell XPS 17 which runs Linux, specificlly Pop!_OS, but also hosts multiple virtual machines, covering Windows and other Linux Distributions.

The full specification can be found below:

Dell XPS 17 (9710)
Intel Core i9-11900HK 2.6GHz Base / 5.0GHz Boost (8C/16T)
64GB DDR4 3200MHz RAM
1TB M.2 PCI-e NVM-e SSD
Intel UHD Graphics 750
NVIDIA GeForce RTX 3060 Ti 6GB GDDR6 (70W)
17-inch Touch Display (3840x2400, 500nits)

At home, the desktop PC and laptops connect to a 49-inch Super Ultra-Wide monitor, specifically the Samsung C49RG90.

The full specification of the monitor can be found below.

Model: Samsung C49RG90
Panel Size: 49-inch
Panel Type: VA
Panel Curvature: 1800R
Aspect Ratio: 32:9
Resolution: 5120x1440
Refresh Rate: 120Hz
Variable Refresh Rate: AMD FreeSync 2 (48-120Hz Range)
Response Time: 4ms (GTG)
Colour Accuracy: 125% sRGB, 92% Adobe RGB, 95% DCI-P3
High Dynamic Range: HDR1000
Brightness: 600cd/m2 (Typical), 1000cd/m2 (Peak)

My workspace is kept fairly minimal, the Dell XPS 17 and Apple MacBook Pro connect to the monitor directly (DisplayPort v1.4), with peripherals connecting via a UGREEN 5-in-1 USB-C Hub, which simplifies the cable management.

As you can see, for peripherals, I primarily use the Keychron K1 Version 4, Logitech MX Master 3 and Logitech Brio Ultra HD Pro. The keyboard and mouse were selected as they include “easy-switch”, which makes it simple to toggle between multiple Bluetooth devices.

The photo also highlights my microphone setup, which is a Blue Yeti connected to the Blue Yeticaster, which includes the Compass Boom Arm and Radius III Custom Shockmount. The microphone is primarily used for video conferencing, screencasting, webinars and podcasts.

The photo below provides a closer look at the Samsung C49RG90 monitor.

As highlighted in the photo below, my desktop PC is hidden under the desk, with the case door exposed providing easy access to the components.

The desk and matching pedestals are from Three Counties Office Furniture. They are designed for corporate use, therefore are hard-wearing and include integrated cable management.

Finally, I use a Herman Miller Mirra 2 chair, which aims to balance comfort and personalised ergonomics in a single design.

Home Server

Alongside my desktop PC and laptops, I have a small home server, which is connected to my Samsung Q7FN QLED TV. The server manages local and cloud backups, media streaming, and collaboration (video conferencing).

I selected an Apple Mac mini (NOV-2018) for the server, thanks to its excellent power-efficiency (150W max), high-performance I/O (4x Thunderbolt 3, 2x USB-A 3.0) and small form factor. The Mac mini also offers versatile video conferencing capabilities covering Apple FaceTime, Zoom, Microsoft Teams, Google Meet, Cisco WebEx, WhatsApp Video, etc.

The full specification of the Mac mini can be found below:

Apple Mac mini (NOV-2018)
Intel Core i5-8500B 3.0GHz Base / 4.1GHz Boost (6C/6T)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
256GB PCI-e SSD
Intel UHD Graphics 630
2x 500GB Samsung Portable T5 SSD (USB 3.1 Gen2)

The 1256GB of local SSD storage is split across multiple drives for resilience, with automated backups being completed by Carbon Copy Cloner and Cloud Storage via Dropbox.

To support my minimal (hidden-wire) entertainment setup, the Mac mini is located in a cupboard under the stairs, connected directly to the Samsung One Connect box via HDMI.

Work Setup

At work, I use the Apple MacBook Pro and/or Dell XPS 17 (depending on the workload).

I use two 27-inch Dell U2718Q IPS HDR 4K monitors, connected via a StarTech Dual-DisplayPort to ThunderBolt 3 Adapter. The peripherals include a Logitech Craft, Logitech MX Anywhere 3 and Logitech Brio Ultra HD Pro.

The desk is very basic, with a simple dual-monitor stand. Similar to at home, I use a Herman Miller Mirra 2 chair at work.

Pop!_OS

Wed, 30 Jun 2021 00:00:00 -0500

Over the years, I have used a wide range of Linux distributions, including Mandrake, SUSE, Ubuntu, Fedora, etc.

Since 2011, I have favoured Fedora thanks to its speed, reliability and focus on innovation. Fedora 34 is a great example, which was one of the first distributions to include GNOME 40, the first major release of GNOME in the past decade.

The “Four Foundations” of Fedora (listed below) resonate strongly with me, especially the focus on community.

Freedom: We are dedicated to free software and content.
Friends: We are a strong, caring community.
Features: We care about excellent software.
First: We are committed to innovation.

With that said, although an admirable position, the “Freedom” foundation can sometimes act as a barrier when working in a world that is still encumbered by proprietory and/or protected software.

For example, the use of NVIDIA proprietory drivers can be painful to configure and maintain on Fedora, especially in devices that use NVIDIA Optimus (hybrid graphics) technology.

Enter System76 a small US-based company with the goal to sell computers that are pre-installed with Linux. To enable this outcome, System76 developed a Ubuntu-based Linux distribution known as Pop!_OS, which targets “developers, makers, and computer science professionals”.

Knowing that Pop!_OS is engineered alongside x86 hardware sold by System76, it naturally simplifies common pain points, such as driver compatibility, disk encryption, firmware management and application provisioning.

These benefits, alongside the recent release of Pop!_OS 21.04 + Cosmic sparked my curiosity, triggering me to explore an alternative to my beloved Fedora.

As my test system I selected a Dell XPS 15 (9510), with the following specification:

Dell XPS 15 (9510)
Intel Core i9-11900H 2.5GHz Base / 4.9GHz Boost (8C/16T)
64GB DDR4 3200MHz RAM
1TB M.2 PCI-e NVM-e SSD
NVIDIA GeForce RTX 3050 Ti 4GB GDDR6 (45W)
15.6-inch OLED Touch Display (3456x2160, 400nits)

System76 offer two Pop!_OS images, one that includes the proprietary NVIDIA driver preinstalled.

This is the option I selected and confirm it installed perfectly, without any manual intervention or additional configuration. I was even able to update the NVIDIA drivers from the Pop!_Shop.

Anyone coming from Windows or macOS likely won’t appreciate the value in having an image with preinstalled drivers, but installing, optimising and maintaining NVIDIA drivers on Linux can be a nightmare. As highlighted by the video below, Linus Torvalds (creator of Linux) has been fairly vocal on this topic.

I am pleased to report that the drivers not only work seamlessly but are also tightly integrated, with an option to automatically switch between integrated and dedicated graphics, leveraging NVIDIA Optimus technology, as well as the ability to select a specific graphics processor per application.

As Pop!_OS is a fork of Ubuntu, it uses the GNOME desktop environment, but will not be updated to GNOME 40 until version 21.10. Thankfully, this is not a huge deal, as System76 include a lightweight overlay from GNOME, known as Cosmic.

Ordinarily, I would be against any unnecessary customisation of GNOME, however, it is as if Cosmic was designed specifically for my workflow. I would describe it as a cross between GNOME 3.x and macOS, with native settings to customise the Dock, Activities, Workspaces and the Launcher.

Honestly, I think Cosmic is close to perfect! I set the Dock too small and hidden, positioned on the left (I commonly use widescreen displays). I hide the Activities and Workspaces buttons, with the Workspace view accessible from the right.

With these simple changes, which can be configured without any GNOME shell extensions, the desktop is clean, minimal and keyboard-centric (exactly how I like it).

Although it took me a few minutes to adjust to the Advanced Package Tool (I am used to RPM Package Manager with Fedora), I was quickly able to install my common software without any issue.

Thanks to Pop!_OS being Ubuntu-based (dpkg), its software repositories are very well maintained, helped further by the fact that Flatpak is natively integrated into the Pop!_Shop.

In summary, I have been very impressed with Pop!_OS. I have not experienced any issues with the Dell XPS 15 (9510), with the system configured to use “hybrid graphics” by default.

If you are new to the Linux world, have a device with NVIDIA graphics or simply want a simplified Ubuntu-based experience, I would seriously recommend checking out Pop!_OS.

At this time, I plan to continue to use Pop!_OS as my primary Linux distribution (sorry Fedora). It will be interesting to see how long this remains true!

GitHub Copilot

Wed, 30 Jun 2021 00:00:00 -0500

Yesterday, GitHub announced a new capability known as GitHub Copilot, currently available as a Technical Preview.

GitHub Copilot is pitched as an “AI Pair Programmer”, which goes beyond traditional code completion by suggesting full lines of code and even entire functions.

It is powered by Codex, a new AI system created by OpenAI. It uses the docstring, comments, function names, or the code itself to suggest potential code matches.

For example, it is possible to write a comment describing the desired logic, at which point GitHub Copilot will suggest the code itself.

GitHub has confirmed that they leverage publicly accessible code on GitHub to train the AI model.

This is an incredible dataset for training, as GitHub includes a massive range of examples covering almost any scenario, as well as relevant meta-data regarding the popularity of the code, etc.

Therefore, assuming GitHub Copilot works as described, the ramifications of this capability could be profound.

GitHub Copilot should improve developer speed and code quality, as well as reduce the risk of errors and bugs. It could also act as a powerful enabler for new developers, providing real-time support when starting a new project or learning a new language/framework.

An alternative perspective could be that GitHub Copilot would negatively impact the competency of the developer community, making the individuals more reliant on tooling to produce code (similar to capabilities such as spell checking and grammar).

As an optimist, I prefer to consider the positive ramifications and hope that the benefits of GitHub Copilot will allow the developer community to prioritise more complex problems, instead of sweating commodity code.

However, I suspect the most controversial aspect of this release will be the impact on Copywrite law.

For example, it is safe to assume that developers regularly search and copy code from public repositories on GitHub (as well as other sources). In theory, they should be checking and adhering to any associated licenses, but I suspect this is not common practice.

There are frequent examples of open-source code finding itself into proprietary projects without any documented acknowledgement of the source, etc.

However, with GitHub Copilot, the process of copying code has been industrialised, making it easier and therefore more prevalent. For example, if GitHub Copilot suggests code from an open-source project, is the developer ok to simply accept the suggestion? Is this any different to manually searching and copying?

This is likely less of an issue for individual developers working on royalty-free and/or personal projects, but more complicated for businesses that custom develop software.

I am a bigger believer in the open-source community, collaborating and learning from each other to solve increasingly complex problems. However, I also believe the community must be protected to ensure it remains healthy, especially individuals that have invested countless hours to progress a specific domain.

I hope GitHub have considered these points and include relevant controls to ensure that developers and any associated licenses are respected.

Finally, I commend GitHub for continuing to innovate. I see this as the first step towards true AI-assisted development, however, I predict a strong and polarising response from the community!

UPDATE (04-JUL-2021)

The team over at DevOps Directive posted an interesting video of GitHub Copilot completing a range of Leetcode interview questions.

As highlighted by the video, the results were impressive, with GitHub Copilot successfully answering 80% of the questions.

It should be noted that I suspect GitHub Copilot was at least partially trained using classic interview questions, which are commonly well defined and structured. Therefore, I suspect the success rate would drop significantly in scenarios with greater nuance or uncommon constraints.

Either way, the technology looks promising, assuming GitHub can navigate the privacy, legal and copyright challenges, whilst respecting developer contributions that help train the model.

Windows 11 Announcement

Thu, 24 Jun 2021 00:00:00 -0500

Earlier today, Microsoft revealed the “next generation” version of Windows, now officially known as Windows 11.

As predicted, Microsoft updated the brand (Windows 11), alongside the introduction of a new user interface, updated application store and tighter Teams integration for collaboration. There was no mention of the Power Platform to promote low-code development and process automation or any new hardware.

Panos Panay (Chief Product Officer for Windows) showcased the key features of Windows 11, with a focus on the user interface, which essentially matched the experience seen in the recently leaked build.

The video below from CNET provides a seven-minute summary of the event.

The Windows 11 user interface is heavily inspired by Windows X, which includes a new Start menu and an updated Start button that are both centred on the taskbar (but can be aligned to the left in the settings).

Alongside the new Start menu, there are several smaller changes, including improved discoverability for Snap and Virtual Desktops (a welcome addition), as well as widgets that can be accessed via a glass themed side panel.

Finally, Windows 11 has better Windows management for multi-monitor setups and an improved touch experience, including larger touch targets, when running on a “touch only” device.

Overall, the Windows 11 user experience looks to be a positive step forward, however, it does not address the underlying inconsistencies.

For example, Microsoft has only updated the Windows 10/8 era user interface components, leaving the legacy components, which date back to Windows 95/NT, unchanged.

I was hoping for something a little more comprehensive but can appreciate the sheer weight of the task when you consider the legacy Windows must attempt to protect.

Alongside the user interface, there was a big focus on the Application Store, which now supports Win32 applications, alongside Universal Windows Platform (UWP)and Progressive Web Apps (PWA). This change is significant, opening the door to the wider Windows application eco-system, making the store instantly more relevant.

This is a welcome change, especially when combined with the new generous store policy (a clear shot at Apple) that, under certain circumstances, allows developers/publishers to forgo any revenue-sharing fees.

Recognising that the application store is not a big revenue driver for Microsoft, means the positive PR gained through this change will likely outweigh any lost revenue.

The final announcement, linked to the application store, was the ability for Android Apps to be downloaded, installed and ran locally.

This is being achieved through a partnership with Intel and Amazon, with Intel delivering an “Intel Bridge” technology that enables Android applications to run on x86 systems (including AMD), with distribution coming via the Amazon App Store.

The unusual partnership is likely a direct response to Apple/Google, which are both actively enabling compatible mobile applications to run on the associated desktop operating system (e.g. iOS to macOS and Android to ChromeOS).

This capability was not part of the recently leaked build, therefore it is unclear how well the Android Apps will run on Windows. However, the documented setup process (outlined below) appears to be a little cumbersome.

Load Windows Application Store
Download Amazon App Store
Run Amazon App Store
Download Android App
Run Android App

Finally, Microsoft detailed the new minimum specification required to run Windows 11.

Processor: 1GHz Dual-Core 64-bit
Memory: 4GB
Storage: 64GB
Graphics: DirectX 12 (WDDM 2.0)
Display: 9-inch @ 720p
Security: Trusted Platform Module (TPM) v2.0

I suspect the biggest point of controversy will be Trusted Platform Module (TPM) v2.0. First introduced in 2014, TPM v2.0 is an international standard for a secure cryptoprocessor, designed to secure hardware through integrated cryptographic keys.

Therefore, although not guaranteed, any system built after 2014 should be compatible. However, this is a little more complicated for “build your own” desktop systems, which are not commonly equipped with a dedicated TPM module.

In this scenario, the system relies on a compatible processor with a secure enclave. This means Intel 8th Generation or AMD Ryzen 3000 Series.

Unfortunately, with limited innovation from Intel over recent years, I suspect many users still have processors that pre-date the 8th Generation.

In conclusion, I felt the event itself felt quite subdued, the words expressed excitement, but the delivery did not. If Microsoft had hoped to rival the presentation quality and energy of recent pre-recorded Apple events, they fell woefully short.

It is known that Panos Panay enjoys a live audience, which may be impacted his performance.

As for Windows 11, it is an interesting update, but certainly not “game-changing”, with the foundations built firmly on Windows 10. It should be noted that there are some parallels to Windows Vista, with a “glass heavy” user interface, the return of widgets (gadgets) and stricter hardware requirements. Let’s hope this is where the comparison ends!

Finally, it is clear that the Windows “11” branding is not the result of technical innovation, but a need to keep the partner market energised.

Ninja Caves - Update One

Sun, 13 Jun 2021 00:00:00 -0500

Since March, I have been building a platformer game for my son (now six years old), inspired by Crystal Caves, using GameMaker Studio 2.

I have documented my progress in the following articles, including the rationale for the project.

I have continued to make good progress and have now completed the following features, which can also be reviewed in my backlog.

Level Objects, including walls, doors, ladders, fires, etc.
Collectable Objects, including coins, crystals, abilities.
Obstacle Objects, including water, lava, spikes, spinning saw.
Destructible Objects, specifically crates.
Story Objects, specifically signposts.
Player Movement, including keyboard/mouse and controller support.
Player Animations, including idle, run, jump, fall, crouch, climb, glide, hit, death.
Player Weapons, specifically ninja stars.
Player Health, including health bar.
Player Special Abilities, including speed, jump and glide.
Enemy Types, including customisable attributes.
Enemy Movement, including keyboard/mouse and controller.
Enemy Animations, including idle, run, jump, fall, hit, death.
Enemy Types, including customisable attributes.
Enemy Weapons, specifically ninja stars.
Lighting System, immitting light from a specific source.
Particle System, specifically lava particles.
Transition System, animating specific events.
Save Game System, persisting level, coins and kills.
Tile Sets, specifically jungle and caves.
Music and Sound Effects.
Menu and Ending Systems.

The video below highlights these features in action across the first three levels, which start in the jungle, before entering the caves.

The majority of the game will occur in the caves (I am thinking ten levels), with each level challenging the player with unique obstacles, supported by specific abilities.

The game has three basic objectives, collect the coins, find the crystal, progress through the level. The ninja enemies can be defeated using ninja stars, but this is not a requirement.

The video below shows the game running on an Apple iPad Pro (Apple iOS 14.6), including Bluetooth controller support.

It is worth noting that Apple iOS assigns the controller to slot 1 (not the default slot 0), which caused me some initial issues, as I had naively hardcoded slot 0 throughout my code.

As part of the most recent release, I also did a lot of clean-up across my codebase, which is available for review (and reuse) on GitHub.

I am particularly pleased with my lighting system, which uses a simple surface overlay and cutout object to deliver the illusion of light immitting from a specific source. The code assigned to the draw event can be found below.

/// @desc Draw Lighting

// Surface Overlay

if (surface_exists(lighting_surface) == false) {
lighting_surface = surface_create(room_width, room_height);
}

surface_set_target(lighting_surface);

draw_clear_alpha(c_black, tansparency); // Set Room Colour

// Lighting Cutout

with(oLightingCutout) {

    var wobble_amount_x = image_xscale + random_range(-wobble, wobble);
    var wobble_amount_y = image_yscale + random_range(-wobble, wobble);

    gpu_set_blendmode(bm_subtract);
    draw_sprite_ext(sprite_index, image_index, x, y, wobble_amount_x, wobble_amount_y, 0, c_white, 1); // Default Cutout Lighting

    gpu_set_blendmode(bm_add);

    draw_sprite_ext(sprite_index, image_index, x, y, wobble_amount_x, wobble_amount_y, 0, colour, intensity); // Custom Cutout Lighting

    gpu_set_blendmode(bm_normal);
}

The code is split into two parts, the first creates the surface overlay and applies transparency based on a room variable. The second leverages a blend mode to remove the surface overlay for the “oLightingcutout” object.

It should be noted that the surface overlay must be the power of two (e.g. 1024, 2048, 4096) to ensure compatibility with Apple iOS/iPadOS devices.

The image below shows the “oLightingCutout” objects applied to the room.

In a future release, I intend to transition my player object (oPlayer) to a finite state machine, which is a method of programming that uses sequential “states” to perform tasks in a logical form. This approach will hopefully streamline the player object, which has become cumbersome due to the number of animations.

The main challenge I have faced is balancing the difficulty, recognising the target audience (a six-year-old). I hope to make the game challenging, but still fun for someone playing their first 2D platformer.

I will continue to post periodic development updates. The game videos, backlog and source code can be found linked below.

Assuming the game maintains a reasonable level of quality, I plan to publish for PC, macOS, Linux, iPadOS, and AppleTV (controller support required).

Journey to the Cloud

Thu, 10 Jun 2021 00:00:00 -0500

Although I post frequently about technology topics that relate to my work, I rarely post on my personal blog about specific work accomplishments.

With that said, the past two years have been unlike anything I have experienced before, with a corporate separation and a major acquisition, all successfully delivered during an unprecedented global pandemic.

Chris Keeley (Elanco CIO) and I recently shared a few highlights from this remarkable journey via a couple of blog posts.

Recognising that these achievements were delivered virtually (due to the pandemic) is evidence of the agility provided by modern architecture, as well as the dedication and resilience of a phenomenal team!

The published image below makes me smile, as it includes my daughter joining an IT Townhall, during one of the many “childcare + work” days.

Over the coming months, we plan to share more details regarding this journey, therefore be sure to follow us at: Blog, LinkedIn, Twitter, Facebook, Instagram.

Windows 11

Thu, 03 Jun 2021 00:00:00 -0500

Yesterday, Microsoft announced an event scheduled for 24th June to discuss the “next generation” of Windows.

This event was previously teased by Satya Nadella (Microsoft CEO) at Build 2021, where he said…

“Soon we will share one of the most significant updates to Windows of the past decade to unlock greater economic opportunity for developers and creators.”

Following the announcement, there has been a lot of online speculation (and confusion) as to what Microsoft intend to reveal.

When Microsoft released Windows 10 in 2015, it was unofficially described as “the last version of Windows”, linked to the “Windows-as-a-Service” release methodology.

Although Microsoft has never formally acknowledged this positioning, they have also done very little to dispute it. Even Cortana, the Microsoft Virtual Assistant, agreed.

In reality, Microsoft was always going to release a “new” version of Windows, as Windows is a powerful brand that has been used over the years to reinvigorate PC sales. This is especially important for Microsoft partners (e.g. Dell, HP, Lenovo, etc.)

With this in mind, what do I expect from the event? Will this be the most significant update to Windows in the past decade?

My prediction, no…

If anyone is expecting a reimagining of Windows and/or new foundational architecture, prepare to be disappointed. For example, although I would love to see Microsoft aggressively target an immutable operating system or container-based application architecture, the need to maintain legacy hardware/software compatibility continues to be a significant barrier.

As a result, I expect to see the following headlines:

New Brand (Windows 11)
Updated User Interface
Updated Application Store
Enhanced Collaboration Capabilities (Teams)
Enhanced Low-Code Development Capabilities (Power Apps)
Enhanced Workflow Automation (Power Automate)

The real star of the show will be the new user interface, which I expect will be more “touch-friendly” and take inspiration from the recently cancelled Windows X project.

All other headline features are a continuation of existing initiatives previously planned for Windows 10, which will now fall under the new brand (Windows 11).

To be clear, this release will still be Windows 10 under the hood, with no significant changes to the foundational architecture.

As a technologist, this prediction feels underwhelming. However, I can count at least seven layers of “user interface” within Windows 10 (listed below). Therefore, achieving some level of consistency would be a positive step forward from a user experience standpoint.

Fluent Design (Windows 10 Era)
Metro (Windows 8 Era)
Windows 7 UI
Windows Vista UI
Windows XP UI
Windows 2000 UI
Windows 95/NT 4.0 UI

Unbelievably, there are still traces of the Windows 95/NT user interface within Windows 10, specifically screensaver settings, etc.

Yes, the preview icon is a CRT monitor, last used two decades ago.

Additionally, I hope Microsoft take the opportunity to make key features more discoverable, such as Windows Snap and Virtual Desktops, which are both obscured in Windows 10, resulting in minimal usage.

Finally, I do believe there is a small chance we will see new hardware at the event, potentially the Surface Neo, which is the previously announced dual-screen device that has been “missing in action” in recent months.

Microsoft may use this device to showcase some of the new touch-friendly user experience, also attempting to usher in a new form factor for Microsoft partners.

Ninja Caves

Sun, 30 May 2021 00:00:00 -0500

Earlier this month I completed a personal coding challenge, dedicating thirty minutes a day for thirty days.

I built a platformer game for my son (five years old), inspired by Crystal Caves, using GameMaker Studio 2.

I documented the process across a few blog posts and daily videos, as well as published my code on GitHub (links below).

One of the goals of the challenge was to reignite my passion for writing code, which has taken a backseat throughout the pandemic due to family/work commitments.

As a result, I am pleased to report that following the thirty days, I have continued to develop the game, polishing some of the rough edges and adding new features. I am not coding with the same frequency or intensity, but still making steady progress and enjoying the process.

The video below includes a brief walkthrough of the first level, which includes the new “ninja theme”, as requested by my son.

As highlighted in the video, the following new features have been added.

Updated player and enemy sprites.
Updated tile set.
Updated enemies, including different behaviours.
Updated weapons, including bullets.
Updated objects, including coins, hearts, signs, etc.
Updated on-screen display, including health, equipment, score, etc.
Updated animations, covering player, enemy, explosions, etc.
Improved controller input.
Implemented player health, including health bar.
Implemented sound, including music and effects.

My son has been an avid tester, providing plenty of feedback and ideas. The video below including a few highlights of him playing through the first level, testing the core game mechanics.

The game is currently running on macOS, using the 8BitDo N30pro controller. The controller works well (especially for players with small hands), but the Bluetooth range is limited, hence the cable.

My son has also been eager to contribute, specifically the creation of new levels, features etc. The picture below is a design he created for a cave level, where the floor is lava.

He has also suggested we include a shield (to block enemy ninja stars) and a glider to help the player navigate across the level.

Overall, I have been inspired by the process and therefore have decided to continue developing the game until it reaches a natural conclusion (or my son become bored).

Assuming the game maintains a reasonable level of quality, I will look to publish, likely targetting PC, macOS, Linux and AppleTV (controller support needed). I will also publish periodic development blogs and videos, tracking the progress.

GitHub Linguist Overrides

Thu, 20 May 2021 00:00:00 -0500

Last month I completed a personal coding challenge, where I dedicated thirty minutes a day for thirty days.

I decided to build a platformer game for my son, using GameMaker Studio 2, with the game logic being written in GameMaker Language (GML).

I documented the process in the following three articles:

Whilst using GameMaker Studio 2, I noticed that GitHub detects files with the “.yy” extension as Yacc language files, which stands for “Yet Another Compiler-Compiler”.

Although not a major issue, it is slightly annoying to see a GameMaker Studio 2 projects be reported by GitHub as 98% Yacc. I would prefer JSON or maybe even GML, as they are generated by GameMaker Studio 2.

As a result, I decided to research the linguist library which is used by GitHub to detect blob languages, ignore binary files, suppress generated files in diffs, and generate language breakdown graphs.

Thankfully, the library supports “overrides”, allowing for custom override strategies for language definitions and file paths.

As documented, I created a “.gitattributes” file with the following code:

# Re-classify .yy files
*.yy linguist-language=GML

The “.gitattributes” must be stored in the root of the GameMaker Studio 2 project, similar to how you would use “.gitignore”.

Once pushed to GitHub, you should see the “languages” section update. This took several hours post push, I assume driven by a scheduled job executed by GitHub.

Depending on your preference, you may want to reclassify “.yy” files as JSON or maybe even ignore them completely. These are both viable options using linguist overrides.

PS5 Optical Audio Delay

Mon, 10 May 2021 00:00:00 -0500

In November, I was lucky enough to secure a Sony PlayStation 5.

My current entertainment setup includes a Samsung Q7FN 4K TV and a Bowers & Wilkins MT-50 sound system, connected via a Sony STR-DN1050 AV receiver.

The Samsung TV was purchased in 2018 and supports 4K at 60Hz (4K60), including HDR10+ and Variable Refresh Rate (VRR). Unfortunately, the Sony receiver is a little older, purchased in 2014 and therefore is limited to the HDMI 2.0 standard, not HDMI 2.0a/b or HDMI 2.1.

To ensure I do not lose features such as High Dynamic Range (HDR) and Variable Refresh Rate (VRR), I connect my HDMI 2.0+ devices (such as the PlayStation 4/5 and Apple TV) directly to the Samsung TV, leveraging a digital optical cable for audio, connected to the Sony receiver.

The setup works reasonably well, with the source devices configured to output bitstream audio, delivering an encoded Dolby Digital or DTS 5.1 signal, which can be decoded by the Sony receiver. Although not the highest possible audio quality, it works perfectly well for daily usage, delivering immersive 5.1 surround sound.

Unfortunately, with the release of the PlayStation 5, Sony removed the digital optical port as a dedicated audio output (image from Polygon), forcing me to send the audio signal over HDMI via the Samsung TV to the Sony receiver.

I can understand why Sony made this decision, as digital optical cables have bandwidth limitations, which are not compatible with modern lossless formats. However, I suspect this omission will frustrate many users with “legacy” sound systems, especially those who have invested in an expensive receiver, etc.

In theory, passing the bitstream configured audio via the Samsung TV should work. Unfortunately, in my experience, this setup results in a frustrating audio delay (up to one second), which can not be corrected through software. Regardless of the content (e.g. Movie, Game), an audio delay is very distracting, ruining the overall experience.

I tried every possible software configuration option across the Samsung TV, PlayStation 5 and Sony receiver, but failed to find a viable solution. It is likely the Samsung TV software is adding an unavoidable audio input delay, that could only be resolved by Samsung via a firmware update (very unlikely at this point).

One option would be to replace my old Sony receiver. Modern AV receivers are HDMI 2.1 compliant, however, reports from the first generation HDMI 2.1 chipsets have not been encouraging, resulting in widespread reports of black screen issues when viewing 4K120 and 8K60 signals.

Therefore, it feels like any AV receiver upgrade should be delayed until second-generation HDMI 2.1 chipsets are available, hopefully in late 2021 or early 2022. Additionally, AV receivers are not cheap, costing upwards of £500 for something equivalent to my current Sony STR-DN1050.

With these challenges in mind, I decided to try an audio extractor, which essentially splits the audio and video signals from the source, providing a digital optical port for the audio output. An audio extractor is a physical device, but are relatively small and cheap (approximately £30). However, due to the complexity of the HDMI and Content Protection (HDCP) standards, perfect compatibility is not a guarantee.

I purchased the Ezcoo HDMI 2.0 Audio Extractor, which (in theory) supports 4K 60Hz, 4:4:4, HDCP 2.2, 18Gbps, Dolby Vision HDR, 5.1 Audio Breakout. Although not HDMI 2.1 compliant, the HDMI specification matches my Samsung TV, therefore should be compatible.

With the audio extractor connected to the PlayStation 5 and dip switches 3, 4, and 5 in the “ON” position, I am pleased to report the audio delay issue was instantly resolved, delivering 5.1 sound via Dolby Digital or DTS.

The audio extractor does require dedicated power (Micro USB), which I achieved by connecting the included cable to a spare USB port on the PlayStation 5, which removed the need for an external power supply.

With the hardware connected, the PlayStation 5 audio “HDMI Device Type” must be configured for “AV Amplifier”, with the desired “Number of Channels” (e.g. 5.1 ch) and “Audio Format” (e.g. Dolby or DTS) selected.

In addition to the audio, I can confirm no visible loss in video quality or features when compared against a direct connection from the PlayStation 5 to the Samsung TV. This Includes 4K at 60Hz with HDR enabled.

Overall, I would highly recommend this solution for anyone experiencing an audio delay when passing audio from their source system (e.g. PlayStation 5) to an AV receiver via their TV.

In the future, I will likely explore an AV receiver upgrade, but this simple and cost-effective solution has probably extended the life of my current set-up by a few years.

WinWire Fireside Chat

Fri, 30 Apr 2021 00:00:00 -0500

Earlier today, WinWire Technologies published a few snippets from a recent fireside chat with Chris Keeley, Senior Vice President and CIO (also my boss).

The fireside chat focuses on the importance of improving animal health, creating value through innovative products, and how this leads to solving some of the world’s most pressing issues.

Embedded below are the video snippets.

How does the overarching mission of purpose drive daily activities?

What does purpose-driven innovation and leadership mean to you?

How has technology-driven further innovation within your organisation?

How has your organisation pivoted since the pandemic?

Advice to aspiring business leaders?

Great work from Chris! It is always valuable to hear from different leaders, helping to challenge preconceptions and ensure diversity of thought.

30 Days of Code - MVP

Wed, 28 Apr 2021 00:00:00 -0500

Last month I started a personal coding challenge, dedicating thirty minutes a day for the next thirty days.

I decided to build a platformer game for my son (five years old), inspired by Crystal Caves by Apogee Software (3D Realms).

Earlier this month, I shared an update, covering the first fifteen (non-consecutive) days, accounting for approximately six hours of development.

I selected GameMaker Studio 2 as my development environment, which includes a cross-platform 2D game engine, with the game logic being written in GameMaker Language (GML).

GML is an imperative, dynamically typed language commonly likened to JavaScript and C, which can be exported to run on a wide range of platforms (Windows, macOS, Linux, iOS, Android, PlayStation, Xbox, Switch, etc.)

Popular 2D games such as Spelunky and Hotline Miami were developed using GameMaker Studio.

Minimum Viable Product (MVP)

I have now passed the thirty-day milestone and am pleased to report that I have achieved a minimum viable product.

The video below demonstrates the core mechanics of the game. However, the visuals and level designs are for testing purposes only, using freely available assets from Kenney.

Showcased in the video is a basic menu system, player and enemy movement, pick-ups, story narrative, player and enemy shooting, scoring, destructible objects, enemy types, saving, loading, and end credits.

My son has been helping to test the game, providing feedback on the controls, etc. It has been fun watching him explore the game mechanics on the big screen.

One area that has been a point of concern is shooting, which is currently achieved using the classic “right analogue stick + trigger” combination found in most shooting video games.

However, my son has found this combination difficult to grasp (literally), due to his small hands. Therefore, I am actively exploring two mitigations.

The first was to remove the need to press the trigger to shoot, instead, the shooting action would be automatic, based on the direction of the right analogue stick.

The second option was to try a different controller, specifically the 8BitDo N30pro, which supports Windows, macOS, iOS, Android and Switch.

I can see a combination of the two options being the best outcome.

Development

If you would like to review the thirty days of development history, the YouTube playlist below provides a glimpse into my daily progress.

30 Days of Code - YouTube Playlist

The codebase, which can be viewed on GitHub, should provide a solid foundation for a 2D platformer game.

It includes numerous re-usable functions, which should make the game reasonable extensible, with options for new collectables, enemy types, destructible objects, etc.

I also included a few “secret” game mechanics. For example, the shooter can be used to provide a small amount of vertical population, which allows the player to jump higher and longer. Although not immediately obvious, once discovered, it could provide an interesting option to be used within the level design.

If you would like to build/test my game from the source, I would recommend the following GitHub commit, found in the master branch.

30 Days of Code - MVP Code

You will need a copy of GameMaker Studio 2 to build the game.

Next Steps

Although I have reached the end of my thirty-day challenge, I plan to continue developisng the game. The next phase will be to focus on art, story, and level design, which I can already tell will be more complex than the code itself.

If you are interested, you can view my current backlog and all enhancements on GitHub (develop-x branches).

I will also upload all of my art assets, which I plan to create using Aseprite (Windows, macOS, Linux) and Pixaki (iPadOS).

My son has informed me that he would like the game to be “ninja” themed, based on his current love for Lego Ninjago.

Conclusion

I am pleased with my progress over the past thirty days and the challenge accomplished its goal of forcing me to get “hands-on keyboard”.

As with any coding, I stood on the shoulders of giants, leveraging the amazing work from the community, specifically Kenney, Gamesfromscratch, AdamCYounis, Shaun Spalding, Vimlark, DevDuck, HeartBeast, Ric, and itch.io.

Fedora 34

Tue, 27 Apr 2021 00:00:00 -0500

Earlier today, Red Hat officially released Fedora 34.

I would usually test pre-release versions of Fedora in a virtual machine, but due to other commitments, did not get the time with Fedora 34.

As a result, I decided to jump straight into the deep end by upgrading my primary Linux laptop (Dell XPS 13 - 9300).

On paper, Fedora 34 looks like a reasonable update, with some notable new features:

Linux Kernel 5.11, which brings support for modern hardware capabilities primarily targetting AMD, as well as improved support for USB4 and Thunderbolt.
GNOME 40 desktop environment, which includes an evolved design, which aims to streamline the user workflow.
ZSTD compression for the BTRFS filesystem, enabling faster read, write performance and optimising disk space.
Systemd-oomd as default, providing better resource handling during out-of-memory (OOM) scenarios.
Pipewire replaces PulseAudio, delivering more flexible, higher-performance audio management.
i3 Tiling Manager, which is used to control the appearance and placement of windows in a windowing system. i3 delivers performance improvements, flexibility and extensive keyboard shortcuts, ready for a Fedora Spin.
KDE Plasma v5.21, which includes many new features, as well as support for Wayland and ARM.
The usual array of package and toolchain updates, including Ruby v3.0, Ruby on Rails v6.1, Golang v1.16, etc.

With twenty-five years of operating system upgrades behind me, I still default to a fresh install vs. an in-place upgrade. This approach, although more time-consuming, has historically helped to protect against errors, conflicts and performance degradation.

However, with Fedora 34, I decided to change a habit of a lifetime and attempt the in-place upgrade. I was sceptical, especially when reviewing some of the horror stories on Reddit, but am pleased to report the upgrade from Fedora 33 was seamless.

I suspect the Dell XPS 13 helped, as the hardware is well supported by the Linux community, with no need for third-party graphics drivers, etc.

Although the switch to PipeWire for audio is arguably the most valuable update, the most obvious difference is the introduction of GNOME 40, which is the first major release of GNOME since 2011.

GNOME 40 introduces a new versioning scheme, with a single number being incremented with each biannual release. This replaces the more complex versioning scheme, specifically v3.38.

I first used the GNOME desktop environment in 2001 (v1.4.1) and although I have experimented with other desktop environments such as KDE, I have always found myself returning to GNOME.

GNOME releases have not been without their controversy, but it is what I know best and aligns with my prefered workflow.

The biggest change with GNOME 40 is the shift from a vertical flow, to a horizontal flow, which is immediately noticeable when you click the “Super” key or activate the “Activities Overview”.

When looking to open an application, the “App Launcher” slides up from the bottom, delivering a consolidated “Activities Overview” that provides instant access to Workspaces, the App Grid and Dash.

This evolved experience feels heavily inspired by Mission Control and Launchpad found in Apple macOS (which is not a bad thing).

The other key change is the default location of the Dash, which is now at the bottom and re-organised to show favourite and running applications. I do not personally use the Dash (or Dock with macOS), therefore its location is of little consequence (although I do position it on the left with macOS).

GNOME 40 also includes a host of new default touchpad gestures, including three-finger swipes to toggle the “Activities Overview” and switch Workspaces. It should be noted that these gestures are only available with the Wayland display server, which is the default for GNOME 40 in Fedora 34 unless you have installed third-party Nvidia drivers.

Alongside other polish, including new default applications and an updated settings menu, I have so far been very impressed with GNOME 40. The horizontal flow feels very natural and the less intrusive animations are much faster (prioritising productivity over visual effects).

I have only been using Fedora 34 for a few hours, therefore have not thoroughly tested all of my applications (yet). However, so far, everything appears to be working well.

30 Days of Code - Update

Mon, 05 Apr 2021 00:00:00 -0500

Last month I started a personal coding challenge, dedicating thirty minutes a day for the next thirty days.

I decided to build a platformer video game, inspired by Crystal Caves by Apogee Software (3D Realms).

As a kid (1991 - age seven), I played countless hours of Crystal Caves on PC (MS-DOS), an experience I hope to reproduce for my son.

This article marks the halfway point, specifically fifteen (non-consecutive) days, accounting for approximately six hours of development.

I have successfully implemented the following features.

Created two rooms (levels).
Created wall, player, enemy, water pistol and water objects.
Created collision map for wall, player, enemy, water pistol and water objects.
Implemented player movement, specifically run and jump.
Implemented enemy object health.
Implemented enemy instance variability.
Created tile set, including nineteen tiles.
Created player sprite, including idle, running, jumping and falling animations.
Created enemy sprite, including idle, running, jumping and falling animations.
Created water pistol sprite, including water animation.
Created a room camera, which follows the player object.
Created three background layers with a parallax effect.
Created room transition object.
Created room transition animation.
Implemented game restart.
Implemented controller (PlayStation and Xbox) input.
Created menu, including save and load functionality.

The short video below highlights a quick “playtest” session with my son (five years old).

He tried the core features of the game, as well as provided me with an early insight into potential player issues. For example, due to the size of his hands, he struggled to control the direction of the water pistol (right analogue stick) and fire (right trigger) simultaneously.

I have also been producing short (less than 30 seconds) video demos of my daily progress.

30 Days of Code - YouTube Playlist

Finally, you can also view my backlog and code on GitHub.

My next update will be in approximately fifteen days, where I hope to have the core mechanics of the game complete. At which point I will focus on the visuals, story and level design.

30 Days of Code

Thu, 18 Mar 2021 00:00:00 -0500

The past eighteen months have been fairly brutal, with high-pressure work deadlines and the global pandemic that has forced lockdowns, homeschooling, etc.

These commitments/restrictions have not left must time for personal growth and one of my regrets is losing my “connection to the keyboard”.

In short, I have not written a meaningful line of code in over a year.

As a geek, this situation pains me… It also means that my core technical skills have become a little rusty.

Therefore, I have decided to dedicate thirty minutes a day for the next thirty days (likely mornings or evenings) to write some code and hopefully learn a thing or two.

I have decided to set myself the challenge to build a 2D platformer game, targeting my five-year-old son. I have not written a game in several years and have historically focused on top-down shooters, therefore a platformer should be an interesting challenge.

Considering my target audience, the game will need to be fairly simple, but I plan to take inspiration from classics such as Super Mario World (SNES) and Mega Man X (SNES), as well as games that dominated my childhood such as Crystal Caves (PC) and Alex Kidd (Sega Master System).

I plan to use the cross-platform game engine GameMaker Studio 2, written in GML. The game will be made available for the PC with controller (PlayStation and Xbox) support.

If you would like to follow my progress, I plan to post short status updates (demos) to YouTube.

30 Days of Code - YouTube Playlist

You can also view my backlog and code on GitHub.

Wish me luck! Expect another post in thirty days, which will hopefully include a video of the playable game.

Company Watch List

Thu, 11 Mar 2021 00:00:00 -0600

In the article “Digital Resources” I shared a list of resources that I use to support my role working in Information Technology.

Alongside the many technology blogs, communities, and open-source projects, I also keep a close eye on companies that are creating or applying technology in interesting ways. Therefore, outlined below is my current company watch list.

It should be noted that this is an opinionated list, targeting my areas of interest and focus. For example, you will see specific resources related to the role of a CTO, covering software development, automation, artificial intelligence, etc.

Blobr - API Monetisation
Hasura - GraphQL API Creation
Strapi - API Creation
Datasaur - ML Data Labeling
Fiddler - ML Model Performance Management
Augury - Machine Health Diagnostics
ElectrifAi - Pre-Built ML Models
SurveyAuto - Data Visualisation
Boast - AI R&D Tax Credits
Zudy - No-Code/Low-Code Development
Node-RED - No-Code/Low-Code Development
AppGyver - No-Code/Low-Code Development
Netlify - Web App Hosting
Snyk - Secure CI/CD
ReadMe - Developer Documentation
Contentful - Headless CMS
ButterCMS - Headless CMS
Integromat - Integration and Workflow Automation
Levity - Workflow Automation
UiPath - Robotic Process Automation
Algorithmia - MLOps
Sedai - Infrastructure AIOps
Reflect - Automated Regression Testing
Confluera - Infrastructure Cyber Security
Capacity - AI Helpdesk
invideo - Video Creation
Digma - Video Interactive Overlays
Arc () - Developer Talent Recruitment
Braintrust - Talent Recruitment
Slower - Talent Recruitment
Coded Minds - Developer Education
Toybox Labs - 3D Printing
NUVIA - Silicon Design for HPC
Robust - Robot Software
Spacemaker - Real Estate Collaberation
Monzo - Bank
Babylon Health - Human Health Care
Forward Healthcare - Human Health Care
Casper - Sleep
Hectare - Farming Supply Chain
FitBark - Dog Health Tracking
Oggii - Dog Behaviour Analytics
Whistle - Pet Tracking
Karakuri - Food Production
XFarm - Agriculture Platform
Hopin - Online Events Platform
Chargifi - Desk Management for Hybrid Working
Argo - Self-Driving Technology
Pure Electric - Electric Transport
Wing - Drone Deliveries
Rapanui - Sustainable Clothing
Gymshark - Fitness Brand

It should be noted that I have a connection with many of the companies on this list, either as a customer, partner and/or advisor.

My Setup (Q1 2021)

Sun, 07 Mar 2021 00:00:00 -0600

Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

Home Setup

At home, I switch between four devices (one desktop PC and three laptops).

The full specification can be found below:

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
Nvidia GeForce RTX 3090 FE 24GB GDDR6X
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

When remote or travelling, my daily driver is a Razer Blade 15 Advanced (Late-2019). Similar to my custom-build PC, it is primarily used for productivity, collaboration, software development, photo editing, virtual labs and gaming.

The full specification can be found below:

Razer Blade 15 Advanced (Late-2019)
Intel i7-9750H 2.6GHz Base / 4.5GHz Boost (6C/12T)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
1TB Samsung 970 EVO Plus PCI-e NVMe SSD
Intel UHD Graphics 630
NVIDIA GeForce RTX 2080 Max-Q 8GB GDDR6
15.6-inch 4K OLED Touch DCI-P3 HDR400 Display (3840x2160)

I also have a 16-inch Apple MacBook Pro (NOV-2019), which is a secondary device providing native access to the Apple ecosystem for software development (e.g. Xcode), video editing, etc.

The full specification can be found below:

16-inch MacBook Pro with Retina Display (NOV-2019)
Intel i9-9980HK 2.4GHz Base / 5.0GHz Boost (8C/16T)
32GB 2666MHz DDR4 RAM
1TB PCI-e SSD
Intel UHD Graphics 630
AMD Radeon Pro 5500M 8GB GDDR6
16-inch IPS P3 Display (3072x1920, 500nits)

Although I use Linux across all devices via a virtual machine, I find it useful to have one Linux system running on bare-metal. I use Fedora as my primary Linux distribution, running on a Dell XPS 13, which is a very capable laptop that avoids complexities such as dedicated graphics, which can be problematic with Linux.

The full specification can be found below:

13-inch Dell XPS 13 (9300)
Intel i7-1065G7 1.30GHz Base / 3.9GHz Boost (4C/8T)
16GB 3733MHz LPDDR4x RAM
512GB M.2 PCIe NVMe SSD
Intel Iris Plus Graphics
13.4-inch Display (1920x1200, 500nits)

At home, the desktop PC and laptops connect to a 49-inch Super Ultra-Wide monitor, specifically the Samsung C49RG90.

The full specification of the monitor can be found below.

Model: Samsung C49RG90
Panel Size: 49-inch
Panel Type: VA
Panel Curvature: 1800R
Aspect Ratio: 32:9
Resolution: 5120x1440
Refresh Rate: 120Hz
Variable Refresh Rate: AMD FreeSync 2 (48-120Hz Range)
Response Time: 4ms (GTG)
Colour Accuracy: 125% sRGB, 92% Adobe RGB, 95% DCI-P3
High Dynamic Range: HDR1000
Brightness: 600cd/m2 (Typical), 1000cd/m2 (Peak)

My workspace is kept fairly minimal, the Razer Blade 15 Advanced, Apple MacBook Pro and Dell XPS connect to the monitor directly (DisplayPort v1.4), with peripherals connecting via a UGREEN 5-in-1 USB-C Hub, which simplifies the cable management.

As you can see, for peripherals, I primarily use the Keychron K1 Version 4, Logitech MX Anywhere 3 and Logitech Brio Ultra HD Pro. The keyboard and mouse were selected as they include “easy-switch”, which makes it simple to toggle between multiple Bluetooth devices.

The photo below provides a closer look at the Samsung C49RG90 monitor.

Additional peripherals include headphones (Razer BlackShark V2 and Bang & Olufsen BeoPlay H8), controller (Microsoft Xbox Elite Series 2), flight controller (Saitek X-55), steering wheel (Logitech G29) and virtual reality headset (Oculus Quest).

As highlighted in the photo below, my desktop PC is hidden under the desk, with the case door exposed providing easy access to the components.

The desk and matching pedestals are from Three Counties Office Furniture. They are designed for corporate use, therefore are hard-wearing and include integrated cable management.

Finally, I use a Herman Miller Setu chair, which includes the Kinematic Spine to control resistance, automatically providing support as you recline.

Home Server

The full specification of the Mac mini can be found below:

Apple Mac mini (NOV-2018)
Intel Core i5-8500B 3.0GHz Base / 4.1GHz Boost (6C/6T)
32GB DDR4-2666 RAM
256GB PCI-e SSD
Intel UHD Graphics 630
2x 500GB Samsung Portable T5 SSD (USB 3.1 Gen2)

The 1256GB of local SSD storage is split across multiple drives for resilience, with automated backups being completed by Carbon Copy Cloner and Cloud Storage via Dropbox.

To support my minimal (hidden-wire) entertainment setup, the Mac mini is located in a cupboard under the stairs, connected directly to the Samsung One Connect box via HDMI.

This setup has been especially valuable during the UK pandemic lockdown, allowing my kids to connect quickly, easily and securely with their grandparents and friends on the big screen.

Work Setup

At work, I use the Razer Blade 15 Advanced and/or Apple MacBook Pro (depending on the workload).

I use two 27-inch Dell U2718Q IPS HDR 4K monitors, connected via a StarTech Dual-DisplayPort to ThunderBolt 3 Adapter. To provide consistency, all other peripherals (e.g. Mouse, Keyboard, and Webcam) are the same as home.

The desk is very basic, with a simple dual-monitor stand. Similar to at home, I use a Herman Miller chair at work, however, due to the extended hours, it is the more robust Mirra 2. This chair includes a phenomenal amount of adjustment, ensuring all-day comfort.

Mac Pro Comparison

Fri, 26 Feb 2021 00:00:00 -0600

I recently upgraded my primary PC, which I use for a wide range of purposes, including productivity, collaboration, software development, photo editing, virtual labs, gaming and game development.

The full specification can be found below:

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Corsair Vengeance LPX DDR4 PC4-28800C18 3600MHz RAM
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
Nvidia GeForce RTX 3090 FE 24GB GDDR6X
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

At the time of writing, the total cost for this specification would be approximately £3500 (I paid £2500 for my upgrade). Although not cheap, this combination of a Processor, Memory, Storage and Graphics delivers exceptional performance across a range of workloads (e.g. gaming, development, video editing, machine learning, etc.)

For example, the Nvidia GeForce RTX 3090 is the highest rated graphics card on the market today, powered by the Ampere architecture, delivering market-leading ray tracing and AI performance with enhanced Ray Tracing (RT) Cores, Tensor Cores, and new streaming multiprocessors.

The 3DMark result below highlights the relative global performance of my PC, which outperforms 99% of systems when targeting graphics-intensive workloads. This result indicates that spending more than £3500 on a PC would achieve diminishing returns.

In 2019, Apple released the third generation Mac Pro, which replaced the infamous 2013 cylindrical (trash can) Mac Pro. The 2019 Mac Pro emphasised performance and flexibility, compromising size and power efficiency, which was a positive direction for “pro” users.

Unfortunately, the Mac Pro commands a very premium price point, costing up to £53,448 with all “build to order” options added. Therefore, I was interested to understand how the Mac Pro (equivalent specification) would compare to my PC.

Unfortunately, due to existing commercial partnerships, Apple continues to use Intel processors and AMD graphics across the “pro-line” of systems, which are both outperformed by their primary rivals (AMD processors and Nvidia graphics).

The Intel Xeon W processor remains “stuck” on the legacy 14nm process, which simply can not compete with the 7nm process used by the AMD Ryzen 3000/5000 series.

The same can be said for the AMD Radeon Pro Vega II, which gets annihilated by the Nvidia Geforce RTX 3090, especially with modern workloads focused on ray-tracing, artificial intelligence, etc.

With these limitations, the Mac Pro (specification outlined above) is thoroughly outperformed by my PC, but also significantly more expensive.

The Mac Pro would cost an additional £10,299, which is ridiculous. To put this price into perspective, you could build four equivalent Windows/Linux-based systems for the price of one Mac Pro.

The only “unique” hardware capability exclusive to the Mac Pro is the Apple Afterburner PCI-e card, which accelerates the decoding of ProRes and ProRes RAW video codecs. However, this card is a “build to order” extra, costing £2,000.

I love the Mac, macOS and the Mac ecosystem, but it is impossible to recommend the Mac Pro. I see a lot of creatives (mostly on YouTube) using a Mac Pro, which for 99.99% of use cases makes no sense.

Even if the workload demands Apple proprietary software, such as Final Cut Pro or Logic Pro X, the return on investment of a Mac Pro could only be realised in very specific scenarios, where extreme performance against time-sensitive deadlines are a requirement. All other Mac-specific workloads would likely be better serviced by an iMac and/or the new Mac mini (powered by Apple Silicon).

With this in mind, I would recommend (even die-hard Mac fans) explore transitioning common workloads to leverage open-source and/or cross-platform software (e.g. Windows, Linux, macOS). This approach ensures maximum flexibility and delivers an unbeatable return on investment.

Digital Resources

Sun, 21 Feb 2021 00:00:00 -0600

This article lists the resources I use to support my role working in Information Technology (I have not included industry-specific resources).

Although not exhaustive, I hope these resources can help motivate, inspire, and/or educate. It should be noted that this is an opinionated list, targeting my areas of interest and focus. For example, you will see specific resources related to the role of a CTO, JavaScript, Kubernetes, etc.

I start with some of my resources (LifeinTECH), followed by third-party communities, web resources, curated lists, and podcasts.

LifeinTECH:

Communities:

Web Resources:

Curated Lists:

Podcasts:

Although not immediately obvious from the list, I leverage YouTube, Twitter (see who I am following), Reddit, and GitHub a lot, consuming and contributing to a wide range of community initiatives and open-source projects.

GeForce RTX 3090

Mon, 08 Feb 2021 00:00:00 -0600

Earlier this month, I upgraded my primary Windows PC, which I documented in the following two articles:

I did not immediately upgrade my old Nvidia GeForce RTX 2080 Ti. The Nvidia GeForce RTX 3000 Series graphics cards look great, however, with COVID-19 supply chain issues and the current cryptocurrency boom, availability and pricing have been a nightmare (with no end in sight).

For example, the Nvidia GeForce RTX 3080 Founders Edition (FE) has an RRP of £649.00, but can rarely be found for less than £1500.00. The same can be said for AIB graphics cards, with the cheapest GeForce RTX 3080 RRP being £899.99 (with no availability).

With that said, if you have the right network (and a bit of luck), it is possible to secure these cards at RRP.

I recently purchased the Nvidia GeForce RTX 3090 Founders Edition (FE), which at RRP costs £1399.00, but is frequently sold for more than £2000.00.

It should be noted, although positioned as such, I do not consider the GeForce RTX 3090 a pure gaming graphics card. Nvidia calls it a “Big Ferocious GPU (BFGPU)”, featuring 24GB GDDR6X memory, which makes it equivalent to a “TITAN-class” graphics cards, used for professional workloads (e.g. Machine Learning, Video Editing, etc.)

The GeForce RTX 3090 FE is a monstrous graphics card in every sense of the word. The specification pushes every limit and the dimensions of the card are like nothing I have worked with before.

A summary of the GeForce RTX 3090 FE specification can be found below:

Architecture Family: Ampere
Architecture: GA102
Process Technology: Samsung 8N
Transistors: 28.3 Billion
Streaming Multiprocessors: 82
NVIDIA CUDA Cores: 10496
Tensor Cores: 328
RT Cores: 82
Base Clock: 1.4GHz
Boost Clock: 1.7GHz
Memory: 24GB GDDR6X
Memory Speed: 19.5Gbps
Memory Interface: 384bit
Maximum Digital Resolution: 7680x4320 (8K)
Display Connectors: HDMI 2.1 and 3x DisplayPort 1.4a
Power Requirement: 350W (Recommended 750W PSU)
Power Connectors: 1x 12-pin (Equal to 2x PCI-e 8-pin)

The GeForce RTX 3090 pushes the GA102 chip to its limit, utilising 82 of the potential 84 Streaming Multiprocessors, which is up from 68 on the GeForce RTX 3080 (also based on GA102).

The image below highlights the architecture.

The dimensions of the graphics card are equality impressive, requiring three full slots. The specific dimensions and weight can be found below.

Length: 313mm (12.3-inch)
Width: 138mm (5.4-inch)
Weight: 2189g (4.82pounds)

I always considered my GeForce RTX 2080 Ti a large graphics card, but the GeForce RTX 3090 FE makes it look small.

I have included a selection of photos below highlighting the GeForce RTX 3090 FE.

The first photo shows the retail box, which like other Nvidia products, feels high quality and delivers an excellent “unboxing” experience.

With the lid removed, the box essentially acts as a presentation stand for the graphics card. You can also see the rear 110mm fan, which has the fins angled to pull air through the radiator.

With the graphics card unboxed, you can see the front 110mm fan, which pushes air out of the rear three-slot exhaust.

The photo below shows a closer view of the rear three-slot exhaust and fan, which are larger than anything I have seen before on a graphics card.

The GeForce RTX 3090 FE includes a unique cooling design, which aims keeps noise to a minimum (48.9 dB under load) by leveraging large fans to exhaust hot air through the rear of the case.

This approach works well with my setup, as I have positioned my fans to pull cool air from the front, running over the system components, before being push out of the rear.

The only watch-out is the internal fan on the GeForce RTX 3090 FE, which will likely blow warmer air over the CPU heatsink. It will be interesting to see if this design has an impact on my CPU temperature.

The GeForce RTX 3090 FE includes a new 12-pin power connector, which is smaller and more convenient than the old two 8-pin connectors. However, at this time, very few power supply units include this connector, therefore an adapter required (not ideal).

Thankfully, Nvidia does include the adapter in the box, therefore any (good quality) 750W power supply unit with two 8-pin connectors should be fully compatible.

The final photo shows the GeForce RTX 3090 FE installed, highlighting the size and unsightly power adopter.

I am pleased to report that my Phanteks Enthoo Evolv ATX Mid Tower Case continues to serve me well, with plenty of room to support the large dimensions of the GeForce RTX 3090 FE.

Regarding performance, the GeForce RTX 3090 FE is unsurprisingly a beast, delivering a significant performance boost over my old GeForce RTX 2080 Ti.

I haven’t had the opportunity to test every workload, but I suspect with this much power, the GeForce RTX 3090 FE will deliver previously unseen performance. Although not the best test (due to the low resolution), the screenshot below highlights my 3DMark Time Spy result.

This score compares well against equivalent systems in the UK, ranking within the top 50% (just over average).

Recognising these are 24x7 stable settings, running with air cooling, without any extreme overclocking, custom bios or voltage modifications, I am pleased with the results.

Over the coming weeks, I plan to put the GeForce RTX 3090 through its paces across a range of workloads, specifically applications that can showcase the raw horsepower (streaming multiprocessors) and utilise the 24GB GDDR6X memory.

Ryzen Build - Update

Tue, 02 Feb 2021 00:00:00 -0600

In my previous article “Ryzen Build”, I detailed the specification of my pending PC upgrade, which will replace my ageing Intel i7-6700K from 2016.

This article will focus on the build itself and my initial testing.

Specification

As a reminder, the full specification of my new PC can be found below.

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Patriot Viper Steel DDR4-3600 (18-20-20-38)
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

Thankfully, I was able to reuse a lot of components from my existing PC, specifically the case, power supply unit and graphics card. Not only did this reduce the cost of the upgrade (approximately £1000), but also simplified the build process, as the power supply unit and I/O cables were already connected, routed, and mounted.

Build

I started the build process by installing the AMD Ryzen 3950X processor, Patriot Viper Steel memory, Samsung 980 Pro SSD and the Samsung PM981 SSD. This included the two SSD heatsinks, supplied with the MSI MAG X570 Tomahawk WiFi motherboard.

The photo below shows the Samsung 980 Pro SSD installed before the heatsink was mounted.

The Samsung 980 Pro SSD delivers impressive performance, with up to 7,000MB/s read and 5,100MB/s write.

Once complete, I mounted the motherboard in the case and connected the EVGA SuperNova P2 1000W power supply unit. The MSI MAG X570 Tomahawk WiFi motherboard requires the standard 24-pin ATX power connector, 8-pin auxiliary connector, as well as an additional 4-pin connector.

Thankfully I had all of the required power supply unit modular cables, even though my existing Intel i7-6700K PC did not require the additional 4-pin connector.

The photo below shows the processor and memory, with the Noctua NH-D15 AM4 mounting bracket ready for the heatsink. Recognising the size of the Noctua NH-D15 cooler, this is likely the last photo that delivers a clear view of the installed processor and memory.

The next phase of the build is always the most nerve-racking, applying the thermal paste and installing the Noctua NH-D15 cooler. In reality, testing has shown that if you are using a processor with an integrated heat spreader (like the AMD Ryzen 3950X), the risk of incorrectly applying thermal paste is relatively low. Therefore, I proceeded with the “pea-sized blob” method, which has served me well in the past.

I did not use the included Noctua NT-H1 thermal paste, instead, I selected Thermal Grizzly Kryonaut High-Performance Thermal Paste. This is a personal preference, which I suspect has minimal real-world impact on temperatures.

Once complete, I mounted the massive Noctua NH-D15 cooler.

Next, I installed the PNY GeForce RTX 2080 Ti graphics card, which due to the size and weight has visible flex when the case is stood vertically (likely occurred over the past two-years of usage).

Finally, I connected and validated all motherboard connections, specifically the power supply unit, fans, I/O, etc.

Overall, I am pleased with the build, which is relatively clean thanks to the majority of the cables being routed via the back-panel of the case.

Regarding cooling, the air is drawn in from the front of the case, running over the motherboard, processor and graphics card, before being pushed out of the rear.

Testing

With the build complete, I switched my focus to testing, specifically optimising the BIOS settings, processor and memory frequencies, as well as the memory timings (which for Ryzen-based systems can have a significant impact on performance).

Unfortunately, I immediately encountered an issue with the Patriot Viper Steel memory, which refused to POST with the XMP profile enabled (either via the XMP option or manually defined).

I knew this was a risk when purchasing the memory, which was not officially “approved” by MSI. Memory compatibility is a real lottery with motherboard manufacturers, with each manufacture configuring the sub-timings and voltages slightly differently, which can impact performance and stability.

This issue forced an undesirable amount of testing, requiring systematically testing of memory frequencies, timings, and voltages.

The best I could achieve (stable) with the Patriot Viper Steel memory was 1600MHz (3200MHz) at CL 16 / tRCD 20 / tRP 20 / tRAS 36 / tRC 74 / CR 1T. No amount of additional voltage (memory and/or controller) achieved improved stability.

I had already confirmed the FCLK (Infinity Fabric) was stable at 1800MHz, therefore I was disappointed not to be able to raise the memory frequency above 1600MHz. The tighter timings did help offset the performance impact, but it was still far from ideal (and below the manufacturer specification of the memory).

With these memory settings and a 1:1 Memory and FCLK frequency, I achieved the following benchmark results.

Although the results are certainly acceptable, I was disappointed by the memory compatibility, therefore I decided to RMA the Patriot Viper Steel memory, replacing it with the more expensive (+ £70) Corsair Vengeance LPX 64GB DDR4 PC4-28800C18 3600MHz memory.

On paper, the Corsair Vengeance LPX is identical to the Patriot Viper Steel, with a 1800MHz (3600MHz) frequency and CL 18 timings. However, the Corsair Vengeance LPX is officially approved by MSI.

Thankfully, by disconnecting one fan from the Noctua NH-D15 cooler, the memory could be swapped without removing the heatsink (which would have been a real pain).

With the Corsair Vengeance LPX memory installed, I cleared the BIOS and selected the XMP profile via the XMP option. The system booted into Windows without any errors, which after several hours or testing with the Patriot Viper Steel memory was a positive result.

Finally, I configured my 24x7 settings and re-ran the suite of benchmarks, which validated the system stability and achieved the following performance results.

Overall, I am pleased with the build and performance, which has delivered excellent improvements over my old Intel i7-6700K PC.

Unsurprisingly, the biggest gains can be seen across applications that favour multi-threading, allowing the 16 Core / 32 Thread AMD Ryzen 3950X to demonstrate its potential.

Conclusion

In conclusion, my new PC has easily met my expectations, delivering a significant performance boost across my common workloads.

The memory testing was both frustrating and eye-opening, highlighting the importance of motherboard and memory compatibility, which can have a significant impacted on performance and stability (even with comparable memory).

Looking ahead, I plan to watch the graphics card market closely, hoping to see the Nvidia GeForce RTX 3000 Series stock and prices stabilise. At which point, I will likely explore an upgrade, helping to deliver an additional performance boost across gaming, machine learning, and video editing workloads.

Apple M1 Management

Wed, 27 Jan 2021 00:00:00 -0600

Over the past couple of months, I have been testing the new Apple MacBook Air, equipped with the Apple M1 System on a Chip (SoC).

The three articles linked below document my initial thoughts and findings.

Alongside performance, compatibility and battery life, I have been testing enterprise management, looking to ensure that Apple Silicon equipped Mac’s can be enrolled and managed alongside their Intel counterparts.

The screenshots below highlight the enterprise enrollment experience using Microsoft Endpoint Manager (AKA Microsoft Intune) running on macOS 11 Big Sur. I also tested Microsoft Defender ATP for Mac.

Enrollment is achieved is via Microsoft Company Portal, which can be downloaded from the Microsoft website.

Similar to Apple iOS and Google Android, Microsoft Company Portal is a standalone application that is used to enable the following capabilities:

Device enrollment to access business resources, including Office 365, Productivity, Collaboration, and data (via OneDrive).
Configure Single Sign-On (SSO) to simplify authentication.
Access business resources via company-issued certificates.
Browse and install approved business applications.
View and manage enrolled devices, including the ability to remote erase, etc.
Links to key business contacts for support, etc.

At the time of writing, the Company Portal application for Mac is Intel-only (x86 architecture), therefore runs via Rosetta 2 emulation on Apple Silicon.

Once installed, the user must log in with a Microsoft Account (Microsoft User Principle Name) that is connected to a corporate environment, which will automatically check the device management status.

Microsoft has done a nice job making user privacy transparent, delivered in an easy to consume format. Assuming the user agrees with the privacy statements, the enrollment process will begin.

Similar to Apple iOS and iPadOS, a macOS management profile needs to be installed, which includes the pre-defined configuration setup using Microsoft Endpoint Manager.

Once complete, the device is checked for compliance. If there are any gaps (e.g. Password Policy, Encryption), the user is prompted to make the required updates.

Finally, the enterprise enrollment process will install any pre-defined business-critical software. I configured Microsoft Defender ATP for Mac, which is also running via Rosetta 2 emulation on Apple Silicon.

In conclusion, I am pleased to report that everything appears to be working as designed, which highlights (again) just how impressive Rosetta 2 is at emulating x86 architecture applications on Apple Silicon.

The enrolment process was seamless and the Microsoft Company Portal and Microsoft Defender ATP applications did not any negatives running via emulation. I also tested a full remote-wipe, which initiated instantly, forcing the Mac to restart and prompt for a remote recovery (reinstall macOS 11 Big Sur).

Overall, I am pleased to see that the transition to Apple Silicon does not impact the ability to enable device management within an enterprise.

Ryzen Build

Tue, 26 Jan 2021 00:00:00 -0600

The last time I completed a major upgrade of my primary PC was 2016 when I switched to Intel’s Skylake architecture. I documented the build process and subsequent upgrades in the following three articles.

The full specification of my current (soon to be old) PC can be found below:

Asus Maximus VIII Hero Alpha (Intel Z170)
Intel Core i7-6700K 4.0GHz @ 4.6GHz (4C/8T)
Noctua NH-D15 CPU Cooler
32GB Corsair Vengeance LPX DDR4 PC4-24000C15 (15-17-17-35)
1TB Samsung PM981 M.2 PCI-e NVMe SSD
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

Since the launch of Skylake, Intel has provided little incentive to upgrade. For example, the current 11th generation processors are still based on the underlying Skylake architecture. Therefore, I have been playing the long game, waiting for more transformative technologies such as PCI-e 4.0 to become mainstream.

In 2019, with the release of the AMD Zen 2 (Ryzen 3000 Series) architecture, things started to get interesting! For the first time in over two decades, AMD surpassed Intel by delivering an architecture with competitive instructions per clock (IPC) performance, alongside a significant jump in the number of cores and threads available across their consumer range (up to 16 Cores / 32 Threads).

Thanks to the benefits of the 7 nanometer MOSFET node from TSMC, the Zen 2 architecture was also aggressively priced, undercutting the competition by a significant margin.

In 2020, AMD built on the success of Zen 2 with the launch of Zen 3 (Ryzen 5000 Series), which maintained the 7 nanometers process but delivered other optimisations that resulted in an impressive 15% to 19% performance improvement.

With the release of Zen 3, combined with PCI-e 4.0 going mainstream, I felt the time was right to explore an upgrade of my primary PC.

Requirements

Working in IT, I utilise my PC for a wide range of different workloads, including productivity, collaboration, software development, photo editing, video editing, video effects, virtual labs, gaming and game development.

As my primary PC, my goal has always been to build a high-performance system that balances these workloads. For example, if the system was used exclusively for gaming, I would target a lower core-count CPU with a higher clock frequency. However, with video editing and virtual labs, a high core-count can make a big difference.

With these workloads in mind, I have defined the following hardware requirements:

Respectable VRM Thermal Performance
Minimum 12 Core / 24 Thread Processor
Minimum 64GB DDR4 3600MHz Memory
1x PCI-e 4.0 x16 Slot with Three Slot Clearence
2x M.2 Slots (PCI-e 4.0 x4) for Storage
Intel Wireless Controller compatible with Wi-Fi 6 (802.11ax)
Minimum 1x USB 3.2 Gen 2 (10Gb/s) Type-C
Minimum 2x USB 3.2 Gen 2 (10Gb/s) Type-A

Although some of the requirements are fairly self-explanatory, a few require additional context.

VRM thermal performance is important when targeting a processor with a high core-count, especially if overclocked.
The 12 Core / 24 Thread processor guarantees a significant multi-core performance improvement over my current Intel Core i7-6700K (Quad Core / Eight Thread). It also provides head-room for gaming, knowing that the PlayStation 5 and Xbox Series X are both equipped with 8 Core / 16 Thread Zen2 processors.
The three slot clearance for the PCI-e 4.0 x16 slot ensures compatibility with Nvidia 3000 graphics cards (for example, the Nvidia GeForce 3090).
Wi-Fi 6 (802.11ax), ideally integrated into the motherboard, delivers 37% better performance and 75% improved latency over Wi-Fi 5 (802.11ac). My home network is 802.11ax compatible via the Asus RT-AX86U Router.

I am not loyal to any specific hardware manufacture or brand and am not interested in cosmetic features, such as RGB lighting, etc. Therefore, the system can be considered high-performance workhorse, built at the best possible price.

Specification

The specification below highlights my proposed build, with the new components in bold.

MSI MAG X570 Tomahawk WiFi
AMD Ryzen 9 3950X 3.5GHz Base / 4.7GHz Boost (16C/32T)
Noctua NH-D15 CPU Cooler
64GB Patriot Viper Steel DDR4-3600 (18-20-20-38)
1TB Samsung 980 Pro M.2 PCI-e 4.0 NVM-e SSD
1TB Samsung PM981 M.2 PCI-e 3.0 NVM-e SSD
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

Knowing that my current system is built on a strong foundation, I can re-use many components. For example, my 1000W Power Supply Unit (PSU) delivers more than enough power for the new build and the Noctua NH-D15 CPU cooler, which was first released in 2014, remains one of the best air coolers on the market.

Selecting the motherboard was the most difficult part of the build, recognising that there are hundreds of X570 motherboards on the market. Through my analysis, I believe the MSI MAG X570 Tomahawk WiFi delivers the best price to performance, with high-quality components/features at a reasonable £199. The motherboard includes:

4x Memory Slots (Supports 128GB)
1x PCI-e 4.0 x16 Slot
1x PCI-e 4.0 x16 Slot (Supports x4 Mode)
2x PCI-e 3.0 x1 Slots
2x M.2 Slots (PCI-e 4.0 x4) with Heatsinks
6x SATA 6Gb/s Ports
1x Realtek 8125B 2.5Gb/s LAN Controller
Intel Wi-Fi 6 AX200 and Bluetooth 5.1
5x USB 3.2 Gen 2 (10Gb/s) Type-A and Type-C
4x USB 3.2 Gen 1 (5Gb/s) Type-A
6x USB 2.0 Type-A

Looking at the TechSpot X570 VRM analysis, the MSI MAG X570 Tomahawk WiFi consistently delivered top performance. This is an impressive result, but even more so when you consider the competing Gigabyte X570 Aorus Xtreme costs £769.99.

The MSI MAG X570 Tomahawk WiFi is likely a byproduct of the bad press MSI received with their first-generation X570 motherboards, which underperformed, leaving many of their loyal fans disappointed. Therefore, the MSI MAG X570 Tomahawk WiFi can be considered an apology and a gesture of goodwill (which I will gladly accept).

The most controversial part of the build is probably the choice to go with a Ryzen 3950X (Zen 2), instead of the new Ryzen 5000 Series (Zen 3).

This decision was primarily driven by price to performance, as I was able to purchase the Ryzen 3950X for £430, which is nearly half the price of the Ryzen 5950X at £829.99. The performance difference between these two products is between 15% and 19% depending on the workload, making the additional cost difficult to justify.

Regarding the graphics card, I have chosen to retain my Nvidia GeForce RTX 2080 Ti (for now), primarily due to the launch challenges experienced with the Nvidia GeForce RTX 3000 Series, which remain in very short supply and therefore heavily overpriced. The Nvidia GeForce RTX 2080 Ti remains a very powerful graphics card, capable of Ray Tracing, therefore will continue to meet my immediate needs until the market stabilises.

The approximate cost of the upgrade was £1000 (£430 on the CPU), which I feel is reasonable considering the current economic climate. As with all technology investments, the longer you are willing to wait, the more you can save (as new products displace old ones). For example, I could see me waiting for the inevitable Nvidia GeForce RTX 3080 Ti/Super before upgrading my graphics card.

Conclusion

Overall, I am excited to see how my new build will perform!

On paper, the additional twelve cores and twenty-eight threads, alongside 64GB memory and 1TB storage rated at 7000MB/s read should deliver a healthy boost in performance, especially across multi-threaded workloads.

Once the components arrive, I plan to document the build process and share the performance results.

Docker + M1 + ARM64

Tue, 29 Dec 2020 00:00:00 -0600

Over the past few weeks, I have been testing the new Apple MacBook Air, equipped with the Apple M1 System on a Chip (SoC).

The two articles linked below document my initial thoughts and findings.

Docker compatibility is an important part of my workflow, therefore I was to see the release of the public preview earlier this month.

As documented in the article “Docker on Apple M1” all core functionality appears to be working as designed, however, as highlighted by the table below, performance is sub-optimal.

As part of my testing, I used this blog (LifeinTECH) as common workload, which installs and configures Ruby, Gems, Jekyll and Bundler, before building my Jekyll static site, which is approximately 400MB.

Although Docker supports multi-platform images, the official Jekyll Docker image does not include an ARM64 version. Therefore, like all x86 images running on Apple Silicon, the QEMU emulator is used, which I suspect is having a significant impact on performance.

As a result, I set up an arm64 version of the official Jekyll Docker image, which should run natively on the Apple M1.

I ran the build multiple times to help identify any anomalies.

As you can see from the table below, the MacBook Air receives a huge performance boost.

Not only does the MacBook Air now outperform my 16-inch MacBook Pro, but also my Razer Blade 15 Advanced, when running Docker via virtualisation (no WSL). These are impressive numbers that are more in line with what I had expected from the Apple M1.

It is worth noting that Docker running via WSL 2 on Windows 10 still offers the best overall performance, thanks to the fact that Microsoft has included a native Linux Kernel as part of WSL 2. I would love to see Apple offer the same capability, but this is very unlikely based on their tightly controlled ecosystem.

In conclusion, if you plan to use Docker on Apple Silicon, I highly recommend selecting images that include native support for the ARM64 architecture.

Cyberpunk 2077

Thu, 24 Dec 2020 00:00:00 -0600

I consider myself a hobbyist game developer, someone that enjoys creating simple video games, with no pressure, expectation or desire to achieve a financial return.

Although I would love to master advanced 3D game engines such as the Unreal Engine or Unity, the time and effort required would be significant. Therefore, I focus on 2D game engines such as GameMaker Studio 2.

Over the years, I have shared some of my personal projects, alongside general observations covering gaming related topics:

Although I do not personally use 3D game engines, the topic still interests me and I am eager to learn the concepts, even if I do not get time to apply them personally.

As a result, I follow a range of game developers, including “Triple-A” studios such as Valve, Naughty Dog, Santa Monica Studio, Kojima Productions, Rockstar Games, Crytek, CD Projekt Red, etc.

I also enjoy content from small independent developers such as Dennaton Games, Derek Yu, The Cherno and Vimlark, DevDuck, AdamCYounis, and Thomas Brush, that deliver more relatable content for hobbyists.

This leads me to the recent controversy regarding the highly anticipated release of Cyberpunk 2077 from CD Projekt Red, which resulted in Sony pulling the game from the PlayStation Store and offering a refund to all customers.

Cyberpunk 2077

Cyberpunk 2077 was announced in May 2012, with a teaser trailer in 2013, which both pre-dated the launch of the PlayStation 4. Although I suspect development did not ramp-up until after the launch of the Witcher 3 in 2015, it is fair to state that CD Projekt Red has been “planning” this game for over eight years.

With over 500 people working on the REDengine 4, Cyberpunk 2077 is a serious “Triple-A” game that took every opportunity to market the anticipated advancements in graphics, story-telling and artificial intelligence, delivering what many hoped would be the most immersive game ever created.

The marketing included a ridiculous number of product partnerships, ranging from Cyberpunk 2077 branded snacks, furniture, clothes, smartphones, gaming peripherals, and a lot more.

The team also capitalised on their star, Keanu Reeves, who plays the character Johnny Silverhand, delivering a wall-to-wall advertising campaign across television, online and the high-street.

Recognising the technical ambition and marketing hype, the team at CD Projekt Red were no doubt under extreme pressure to deliver and I suspect underestimated the time it would take to deliver a product across multiple platforms and generations (e.g. PlayStation 4/4 Pro/5, Xbox One/One X/S/X and PC). This became evident with continuous delays, that resulted in the team missing their original release window by eight-months.

Unfortunately, even with the additional development/testing time, the launch of Cyberpunk 2077 was a public relations disaster, with a product that was essentially unplayable on the “base” consoles (e.g. PlayStation 4 and Xbox One).

As highlighted by the Digital Foundry analysis (video below), the PlayStation 4 and Xbox One versions of the game run below 720p (variable resolution) and dip to 15fps. These are performance numbers that have not been seen in several generations and certainly not what consumers have come to expect from the era of high-definition, etc.

Alongside the horrendous performance, the game also suffers from countless bugs, including hitching, missing textures, crashes and even half-finished sections of the game. In one hilarious moment, as seen in the Digital Foundry analysis, the game “stalls” for over a minute before resuming (essentially one frame per minute).

The PlayStation 5 and Xbox Series X do run the game better, leveraging their raw horsepower to deliver a passable experience, but even they cannot protect against the countless bugs.

It is not uncommon for games to launch with bugs, especially open-world games, but this is clearly unacceptable and therefore it is not a surprise that Sony felt the need to intervene.

Cyberpunk 2077 PC Version

On a positive note, the PC version of the game demonstrates the potential of Cyberpunk 2077. It is far from perfect, but certainly not the unplayable mess found on the consoles. This highlights that the PC version was likely the lead platform for development, with the team hoping to be able to scale back the game to run on consoles.

The PC version is also a showcase for Ray Tracing, arguably delivering the most ambitious implementation of this demanding technology since Control.

I have been playing Cyberpunk 2077 on the PC and can confirm, when everything is working as designed, Night City is something special. The density and vertical scale, combined with the advantages of Ray Tracing, delivers a truly immersive experience.

Unsurprisingly, the PC requirements are very high (especially with Ray Tracing enabled), therefore, I thought I would share my specification and configuration.

I have been playing Cyberpunk 2077 on the following system.

Asus Maximus VIII Hero Alpha (Intel Z170)
Intel Core i7-6700K 4.0GHz @ 4.6GHz (4C/8T)
32GB Corsair Vengeance LPX DDR4 PC4-24000C15 (15-17-17-35)
1TB Samsung PM981 M.2 PCI-e NVMe SSD
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
Samsung C49RG90 49-inch (5120x1440 @ 120Hz)

It is worth noting that I use a Super Ultra-Wide Monitor, specifically the Samsung C49RG90, which is a 49-inch monitor running at a 5120x1440 resolution, alongside a high refresh rate (120Hz) and support for AMD FreeSync 2.

With this specification in mind, I have configured the following settings to achieve a stable frame-rate above 30FPS with all Ray Tracing features enabled.

Resolution: 5120x1440
Field of View: 80
Film Grain: Off
Chromatic Aberration: Off
Depth of Field: On
Lens Flare: On
Motion Blur: Low
Contrast Shadows: On
Improved Facial Lighting Geometry: On
Anisotropy: 16
Local Shadow Mesh Quality: High
Local Shadow Quality: High
Cascaded Shadows Range: High
Cascaded Shadows Resolution: Medium
Distant Shadows Resolution: High
Volumetric Fog Resolution: High
Volumetric Cloud Quality: Medium
Max Dynamic Decals: Ultra
Screen Space Reflections Quality: High
Subsurface Scattering Quality: High
Ambient Occlusion: Low
Color Precision: Medium
Mirror Quality: High
Level of Detail (LOD): High
Ray-Tracing: On
Ray-Tracing Reflections: On
Ray-Tracing Shadows: On
Ray-Tracing Lighting: Psycho
DLSS: Performance
Dynamic FidelityFX CAS: Off
Static FidelityFX CAS: Off

Overall, depending on the demand of the scene, I see a frame-rate between 30FPS and 50FPS, which thanks to RPG-style of the game (slower pace) and Low Framerate Compensation (LFC) via FreeSync 2, delivers what I consider to be an acceptable experience.

I use the same settings (1920x1080 resolution) on my Razer Blade 15 Advanced, which thanks to the Nvidia GeForce 2080 Max-Q graphics card and Samsung 970 Pro SSD performs well. It does not achieve the same consistent frame-rate as my main PC but remains surprisingly playable.

At this time I am not far enough through the game to comment on the story, but it is clear that the non-playable character artificial intelligence is not the “game-changer” I had hoped it would be. The city feels alive, but the inhabitants of the city feel disjointed, with scripted reactions that do not always match the situation. I had hoped for better, although (again) this likely due to unrealistic expectations, following the hype surrounding the game.

In short, Cyberpunk 2077 on the PC looks great, plays well and certainly demonstrates promise (I am enjoying the game). It is far from perfect, but certainly not the disaster experienced on the consoles.

Conclusion

I suspect many factors contributed to the dismal launch of Cyberpunk 2077, likely compounded by the current economic situation with COVID-19, etc.

With this in mind, I feel the CD Projekt Red senior leadership must take accountability, as I assume they would have ultimately sanctioned the release. I suspect the decision was driven by the mounting commercial pressure associated with the extensive marketing campaign and many co-promotional agreements, which likely included financial penalties for an additional delay.

Therefore, I empathise with the designers, developers and testers, who must have known (and I assume communicated) the game was not ready for release and therefore have had their talent and hard work questioned by factors outside of their control.

Looking forward, I hope CD Project Red endeavour to improve the situation through free updates, patches, etc. In my opinion, they should postpone any new features (e.g. expansions) until the foundations of the game reach an acceptable level.

Unfortunately, based on the current state of the game, this will not be a simple task. Through my analysis, I question if the base consoles (PlayStation 4 and Xbox One) can ever deliver an acceptable experience. For example, the AMD Jaguar architecture (launched in 2013), alongside the use of mechanical hard drives (80MB/s to 100MB/s) are physical architectural bottlenecks that even the most skilled developer would struggle to overcome.

In conclusion, I feel the launch of Cyberpunk 2077 must become a case study (and warning) for all future game releases. The key learning would be to ensure the marketing and hype do not get ahead of the product itself, which ultimately forced a release that disrespected the CD Projekt Red team, disappointed customers, and resulted in considerable brand damage that may take years to recover.

Docker on Apple M1

Sat, 19 Dec 2020 00:00:00 -0600

In the article “Apple M1”, I shared my initial observations of the Apple transition from Intel to Apple Silicon, using the new MacBook Air (Fourth Generation).

In summary, I was incredibly impressed with the Apple M1 System on a Chip (SoC), which delivered excellent performance and power efficiency as well as a viable transition runway via Rosetta 2.

I concluded that this transition was not only the most ambitious change in the history of the Mac but could also have a profound impact on the personal computing industry.

With that said, due to the immaturity of the architecture, there were some limitations. A big one for me was the inability to run Docker, which I use frequently as part of my development workflow.

Earlier in the week, Docker released the first public preview of “Docker Desktop for M1”, enabling native compatibility. Although still in preview, this is a significant milestone, removing one of the primary blockers that would prevent me from using an Apple Silicon powered Mac as my daily driver.

I did not expect to see a public preview in 2020, therefore the Docker development team should be commended for their quick work. As highlighted in the blog post, the team needed to overcome the following engineering challenges:

Migrate from HyperKit to the Virtualization Framework introduced in macOS 11 Big Sur.
Recompile all binaries of Docker Desktop to run natively on Apple Silicon.
Update engineering tools (CI/CD and Testing) to support Apple Silicon.

This article will focus on my initial impressions of Docker on Apple M1, looking to assess reliability and performance across my common workloads. Recognising this is a preview, I do not expect perfection.

Reliability

Once downloaded, the installation was very simple, essentially identical to the regular Docker Desktop for Mac.

I decided to use this blog (LifeinTECH) as my reference point, as I am very familiar with its performance characteristics building via Docker across a wide range of hosts (Windows, Mac, Linux).

As a reminder, LifeinTECH follows the JAMstack web development architecture, using the Jekyll Static Site Generator. It includes a range of custom plugins to support search, tagging and to enable Progressive Web App (PWA) capabilities (offline access, etc.) The project size is approximately 400MB.

I use the official Jekyll Docker image as part of my build, which itself requires Ruby, Ruby Gems and Bundler.

Leveraging Docker Compose, I was able to use my YAML configuration file to install and configure the required runtime dependencies.

Once complete, the Jekyll build service was executed, which generates a static site for production use. As you can see from the image below, the build took approximately 614 seconds to complete (more on performance later).

Everything worked as designed, including the Docker Dashboard GUI.

The Docker Dashboard GUI includes some useful real-time statistics of any running container, highlighting CPU, Memory and Storage and Network utilisation.

Everything built as expected, however, I did receive several warnings related to file conflicts. I do not receive these warning when building for WSL 2 on Windows, therefore I suspect this is related to the filesystem differences of the mounted volumes (e.g. AFS, ext4, NTFS).

Thankfully, the generated static site appears to work as designed.

Performance

With reliability looking good, I switched my attention to performance. The table below outlines the build times leveraging the “default” Docker settings for each device.

I have also added the relative Geekbench CPU scores, highlighting the relative performance of the hardware.

As you can see, the MacBook Air running the Apple M1 is very slow, falling behind my 16-inch MacBook Pro and only narrowly beating my old 12-inch MacBook.

The 12-inch MacBook is running an Intel Core i7-7Y75, which is a 7th Generation dual-core processor, that is passively cooled (no fan). As highlighted by the Geekbench scores, the Apple M1 is (in theory) significantly more powerful than the Intel Core i7-7Y75, highlighting a potential software bottleneck.

If I had to guess, I would highlight the following limitations.

Docker on M1 is pre-release (public preview), therefore likely not optimised for the new architecture.
Although Docker supports multi-platform images, I suspect the majority are still built using the x86 architecture, therefore not native to the Apple Silicon. This is the case for the official Jekyll Docker image, which does not include an ARM64 version. In this scenario, I believe the QEMU emulator is utilised, which would have a significant impact on performance.

It is worth noting that Docker running via WSL 2 on Windows 10 is still the gold standard (outside of native Linux), delivering over seven times the performance of the MacBook Air. This is thanks to the fact that Microsoft includes a native Linux Kernel as part of WSL 2, delivering bare-metal performance (CPU and I/O).

Conclusion

In conclusion, I am pleased to have a compatible version of Docker running on the Apple M1. Following my initial testing, everything appears to be working as designed.

Performance is not great when using x86 Docker images, but this is not a limitation of the hardware, therefore I hope to see things improve over time as the software stack matures, optimising for Apple Silicon.

Over the coming weeks, I plan to test some ARM64 Docker images, which I expect to perform better. In parallel, I am interested in evaluating the capabilities of the Apple Virtualization Framework, specifically targetting client-side virtualisation.

Computer Vision - Update

Fri, 04 Dec 2020 00:00:00 -0600

Over the past few months, I have been testing Computer Vision technologies using an Nvidia Jetson Nano Developer Kit. I covered the framing, use cases, architecture and set up in my previous two articles.

This article will focus on the usage of the Jetson Nano Developer Kit to execute some basic Computer Vision workloads, covering image recognition, object detection and semantic segmentation.

The photo below shows my Jetson Nano Developer Kit, connected to my monitor running picture-in-picture via HDMI.

To get started, I used the Inference and Realtime DNN Vision library from Dustin Franklin, which includes a range of example projects, as well as links to the Nvidia Jetson AI Courses and Certifications.

The video below covers the software configuration process, including the setup of Docker and installation of the pre-trained models.

The library comes pre-loaded with forty models, which can be managed and installed via the ‘docker/run.sh’ command.

All models are installed locally on the host and automatically mounted to the running Docker container, which makes viewing the output and future modifications much easier.

Once the initial configuration is complete, the examples projects can be loaded, covering Computer Vision primitives, such as imageNet for image recognition, detectNet for object detection, and segNet for semantic segmentation, which all inherit from the shared tensorNet object.

Image Recognition

The first project focuses on image recognition, which uses classification networks that have been trained on large datasets to identify scenes and objects.

Dusty-NV/imageNet

The imageNet object accepts an input image and outputs the classified object name, confidence of the classified object, and the framerate.

Assuming the Docker container is running, the image detection project can be started using the following command.

./imagenet.py csi://0

The “csi://0” command assumes a MIPI-CSI Camera is installed, for example, the Raspberry Pi Camera.

The photos below highlight the image recognition project in action.

I tested a wide range of objects, including many of my son’s toys. The model was surprisingly good at detecting dinosaurs, with varying levels of confidence.

The second photo highlights higher confidence (99%) when viewing a baseball.

The average framerate for image recognition on the Jetson Nano Developer Kit was 68fps.

Object Detection

The second project focuses on object detection, which aims to identify where in the frame various objects are located by extracting their bounding boxes.

Unlike image recognition, object detection networks are capable of detecting multiple different objects per frame.

Dusty-NV/detectNet

The detectNet object accepts an image as input and outputs a list of coordinates of the detected bounding boxes, alongside the classified object name, the confidence of the classified object, and the framerate.

The object detection project can be started using the following command.

./detectnet.py csi://0

The photo below highlights the object detection project in action.

The object detection project pushed the Jetson Nano Developer Kit a lot harder, resulting in 20fps.

Semantic Segmentation

The final project focuses on semantic segmentation, where the classifications occur at the pixel level instead of the entire image. This technique is useful for environmental perception and images with multiple objects per scene.

Dusty-NV/segNet

The segNet object accepts an image as an input and outputs a second image with the per-pixel classification mask overlay.

The semantic segmentation project can be started using the following command.

./segnet.py --network=fcn-resnet18-mhp csi://0

The photo below highlights the semantic segmentation project in action.

I struggled to achieve clear results with this project, likely due to poor lighting and/or a lot of background noise. The Jetson Nano Developer Kit achieved an average of 28fps.

Conclusion

In conclusion, the three examples projects provide a nice foundation covering Computer Vision primitives, such as image recognition, object detection and semantic segmentation.

The Jetson Nano worked well, delivering acceptable performance considering the cost and passive cooling. It did however become very hot when running the object detection project, potentially highlighting the need for a fan.

Overall, I remain very impressed with the Jetson Nano Developer Kit and believe it is an excellent entry-point for anyone interested in exploring Artificial Technology (AI).

Software List 2020

Fri, 20 Nov 2020 00:00:00 -0600

Update - The table was updated in September 2022, including my latest software choices.

Since 2003, I have been a devoted Mac user, favouring their “Pro” line of notebooks (PowerBook/MacBook Pro). In early-2019, driven by multiple factors, I decided to leave the “Cult of Mac”), switching to a Windows PC.

With my new found freedom, throughout 2019 and 2020, I took the opportunity to experiment with multiple operating systems and device types, including Windows 10, Linux and even iPadOS. I was eager to understand the maturity of each platform related to my common workflows, whilst documenting the transition process.

The articles linked below highlight this journey, describing my experience with the different software/hardware and providing insight into specific areas (e.g. development).

A byproduct of this transition process was the need to establish a common software list that could be utilised across different platforms.

The goal was to make the transition between platforms as seamless as possible, minimising disruption through standardisation and native compatibility.

To achieve this outcome, I actively avoided highly-opinionated software that targeted a specific platform, instead prioritising web-based technologies, accessed via the browser and/or technologies such as Electron.

Although “avoid lock-in” became my guiding philosophy, I did make exceptions, specifically for any software that delivered unique value above “open” alternatives. For example, I use Final Cut Pro on macOS for video editing, even though it is not available on Windows or Linux.

The table below highlights my software selection across macOS, Windows, Linux and iPadOS, covering my common workflows.

As you can see, the primary “desktop” platforms (macOS, Windows and Linux) are comprehensively covered, each offering high-quality software that can be used with a low barrier to entry.

Unsurprisingly, iPadOS (due to its inherent limitations) is a more challenging platform. Although it continues to mature with every release, “power user” tasks such as development remain cumbersome, requiring some creative workarounds.

As to which platform is “best”, I do not believe there is a perfect answer, which is why I would advocate the ability to use any of them. At a high-level, I would summarise

macOS: Best hardware/software integration, often resulting in the most high-optimised experience (e.g. better battery life, etc.) The *nix kernel makes POSIX compliant development seamless (at least on Intel-based Mac’s) but often limited by the tightly controlled Apple ecosystem.
Windows: The most “work-friendly” platform, especially utilising applications from the Microsoft ecosystem (e.g. Office 365). With the introduction of Windows Subsystem for Linux 2 (WSL 2), POSIX compliant development is now native, making Windows the most versatile platform.
Linux: Highly customisable and cost-effective, with the widest array of free and open-source software (FOSS). The main barrier (for me) is the lack of native Office 365 software (e.g. Word, PowerPoint, Excel, etc.)

If you are a long-time user of a single platform but have been considering a change. I feel there is no better time, recognising the maturity of the main desktop platforms and prevalence of web-based software.

Apple M1

Wed, 18 Nov 2020 00:00:00 -0600

In June, Apple announced its intent to transition the Mac from Intel to its own custom silicon, known as Apple Silicon, based on the ARM architecture. This transition will establish a common architecture across all Apple products (e.g. iPhone, iPad, Apple Watch, Mac), which should streamline hardware and software development, enabling new customer experiences.

This is not the first time Apple has completed a transition of this type, most recently, the 2005 switch from PowerPC to Intel. I remember this transition well, as I was mid-way through university and reliant on my trusty Apple PowerBook G4 (part of the PowerPC 74xx family). At the time, PowerBook users had been eagerly awaiting the release of the PowerBook G5, only to discover that the range would be replaced by the MacBook, running an Intel Core Duo processor.

Although the PowerPC to Intel transition can be used as a reference point, the market dynamics were very different.

Intel was the dominant player in microprocessors and x86 had established itself as the default instruction set architecture. As a result, the switch to Intel and x86 was a safe bet, aligning with the industry direction and retaining the existing value chain, with a third-party (Intel) being leveraged to design and build silicon for the Mac.

Apple handled the transition reasonably well, primarily thanks to the higher-performance Intel processors and an emulation layer called Rosseta, which allowed legacy PowerPC applications to run on Intel hardware. As with any emulation-layer, the use of Rosseta resulted in a performance penalty, which was a little painful, especially with applications such as Microsoft Office 2004, which long outstayed its welcome. In 2011, Apple removed Rosetta from the operating system, marking the end of the PowerPC to Intel transition.

The transition to custom Apple Silicon is a little different. For example, Apple Silicon is not a known quality on the Mac and although ARM-based architecture is well established for low powered devices (mobiles, tablets, watches), it has not been tested at scale for general-purpose computing (laptops, desktops). Therefore, it will be interesting to see how it competes against x86 across different workloads.

This is also the first time that the Mac will be vertically integrated, positioning Apple as a direct competitor to industry giants such as Intel and AMD.

ARM Revolution

The words “game changer” are overused in technology, however, I believe the shift to Apple Silicon and an ARM-based architecture has the potential to disrupt the entire personal computing industry.

If Apple succeeds and achieves better performance and power efficiency, delivered in a smaller form-factor at a lower cost, the Mac would have a tremendous advantage over the competition (e.g. Windows PC).

Similar to the launch of the original iPhone, competing products (e.g. HP, Lenovo, Dell) would immediately feel obsolete. Worse still, the markets ability to respond would be slow, recognising the heavy reliance on third-parties (Microsoft, Intel, AMD) to unlock innovation. This has been a major challenge over recent years (one of the reasons Apple has decided to transition), recognising that Intel has consistently failed to meet its performance expectations.

With that said, if Apple fails to deliver better performance and power efficiency or inadvertently alienates developers, they have potentially isolated the Mac, reaffirming its market position as a niche product.

As a reference point, Sony attempted a similar transition with the PlayStation 3, with the decision to design and build a custom processor, known as Cell, based on the PowerPC architecture. Unfortunately, developers did not embrace the change, which put the entire brand (and Sony as a company) at risk. Sony had to live with this decision for seven years, before switching back to the x86 architecture for the PlayStation 4 (and PlayStation 5).

In short, I believe the transition to Apple Silicon is the most ambitious change in the history of the Mac and a decisive moment in personal computing.

Therefore, I am excited to share my first observations of a Mac powered by Apple Silicon, specifically the Apple M1 System on a Chip (SoC).

Apple M1

In 2008, Apple announced the acquisition of Palo Alto Semiconductor (P.A. Semi), a fabless semiconductor company, as well as a stake in Imagination Technologies, designer of the PowerVR graphics processors.

This was the first public indication that Apple was becoming serious about custom silicon. Over the next decade, Apple demonstrated its growing competency, introducing custom silicon across its product portfolio (iPod, iPhone, iPad, AppleTV, Apple Watch). They also started to build unique processors and co-processors for the Mac (Apple T2 Chip) and AirPods (Apple W1 Chip), etc.

Most interestingly, Apple was not just competing, but regularly outperforming its competitors (e.g. Qualcomm), with phenomenal jumps in compute, graphics and machine learning performance. However, recognising these achievements were predominantly restricted to mobile and/or specialist devices, there was no consistent way to benchmark against Intel and AMD.

In recent years, this story has started to change, as smartphones, tablets and laptops have converged. The image below, put together by Anandtech, compares Apple vs. Intel regarding performance over time.

Ignoring the actual numbers, the rate of innovation (performance increase over time) is the real story, which is what has made Apple Silicon a viable alternative to x86 for personal computing. Assuming Apple can maintain this rate of innovation, PC manufactures should be very nervous.

This leads us to the first SoC designed specifically for the Mac, known as the Apple M1. A summary of the specification can be found below.

Apple M1 System on a Chip (SoC)
5nm Process
16B Transistors
8-Core Central Processing Unit (4x High-Power, 4x Low-Power)
3.2GHz Max Clock Rate
12MB (High Power), 4MB (Low-Power) L2 Cache
8-Core Graphics Processing Unit
8GB or 16GB 4266 MT/s LPDDR4X SDRAM
16-Core Neural Engine

I do not plan to detail the specifics of the Apple M1 architecture, if you are interested, I would recommend reading the excellent analysis from Anandtech. However, I have included an annotated image of the SoC, which highlights the core components.

The key takeaway is that Apple has chosen to deliver a System on a Chip (SoC), which means all core components are embedded. This approach has many advantages (performance, power-efficiency and space), but removes any ability to upgrade individual components (e.g. Memory, etc.)

It is curious how much space Apple reserved for the 16-Core Neural Engine, a capability that is not commonly found on x86-based systems. I feel this highlights another area of innovation (secret sauce) that might only become evident over time.

Although the “paper specification” is interesting, it cannot be directly compared to traditional x86-based systems from Intel or AMD. For example, core count, clock speeds, memory capacity, memory speed are less relevant when comparing two different architectures, as they are fundamentally different, with unique strengths and weaknesses.

As demonstrated by the iPhone, we have also seen the advantage that comes from vertical integration (hardware and software). For example, the Apple iPhone 12 Pro has 6GB of Memory, compared to the Samsung Galaxy S20 Ultra, which has 16GB. However, the iPhone consistently outperforms its rivals, partly driven by the level of hardware/software optimisation.

In summary, although the architecture and specification of the Apple M1 are certainly intriguing, I am mostly interested in how they translate to real-world performance, power-efficiency and compatibility.

Apple MacBook Air

I purchased an Apple MacBook Air (Fourth Generation) for testing.

The full specification can be found below.

Apple MacBook Air
Apple M1 8‑Core CPU, 7‑Core GPU, 16‑Core Neural Engine
16GB Memory
256GB SSD
2x USB 4 (Thunderbolt Compatible)
13.3-inch P3 IPS Display (2560x1600, 400nits)

A few areas to note, the new MacBook Air, MacBook Pro 13 and Mac Mini all include the same Apple M1 SoC, however, the MacBook Pro 13 ad Mac Mini have a fan (likely producing better-sustained performance).

The MacBook Air I used for testing includes 16GB Memory, but a lower-binned SoC, with a 7-Core GPU (can be configured with 8-Core).

The total price for this specification was £1,199.00, which is comparable to the previous generation Intel MacBook Air.

Like the price, the design of the MacBook Air is essentially identical to the previous generation, with a slightly better display (P3 colour gamut) and support for USB 4, which is compatible with Thunderbolt.

Overall, The MacBook Air is a great passively cooled (no fans) notebook, that targets mainstream consumers.

Performance

Unsurprisingly, following the launch of the Apple M1, thousands of reviews have hit the Internet, many of which focus on the traditional suite of synthetic benchmarks.

If you are interested in the benchmark results, I would advise referencing the following publications:

I did validate some of the benchmark results, compared against my 16-inch MacBook Pro, which includes an 8-Core Intel i9-9980HK processor and (where appropriate) my old 12-inch MacBook (2017), which includes a passively cooled Intel Core i7-7Y75 processor. The first image highlights the Geekbench CPU score of my 16-inch MacBook Pro.

The image below highlights the same benchmark running natively on the MacBook Air.

As you can see, the M1 crushes the Intel i9-9980HK, which is already impressive, but especially considering it is passively cooled (no fan). Unsurprisingly, my old 12-inch MacBook was the worst performing by a wide margin, with a single-core score of 818 and a multi-core score of 1683. Although this is a great result for the M1, it should be noted that synthetic benchmarks only tell part of the story and Geekbench consistently favours Apple hardware.

As a result, I prefer to focus my testing on real-world performance across my common workloads, covering productivity, collaboration, development, photo editing, video editing, etc. The video below provides a narrated overview of my experience.

Overall, I am very impressed by the results, which is great news, considering the Apple launch event made big claims but failed to provide any specifics. The lack of specifics would usually be a cause for concern, but thankfully the real-world performance speaks for itself.

When compared against my 16-inch MacBook Pro (8-Core Intel i9-9980HK, 32GB Memory, AMD Radeon Pro 5500M), I am amazed by how well the passively cooled MacBook Air performs, especially under sustained load, which would usually result in thermal throttling and therefore a loss of performance. The M1 within the MacBook Air does appear to throttle, but far less aggressively than Intel processors and usually only after nine minutes of sustained load, which is more than acceptable for mainstream consumers.

I am also pleased to report that executing common tasks such as productivity and collaboration are very responsive, equivalent to my 16-inch MacBook Pro and certainly outperforming my old 12-inch MacBook which is also passively cooled. In certain scenarios, for example, wake from sleep, the MacBook Air is essentially instant, tangibly quicker than my 16-inch MacBook Pro, demonstrating the advantage of being vertically integrated.

My only frustration is not being able to run my developer workloads, which would have given me a much better indication of the performance. However, due to software limitations (covered under compatibility), applications such as Docker and Client-Side Virtualisation are not currently usable.

Power Efficiency

Unsurprisingly, the ARM architecture excels when it comes to power efficiency.

Recognising the current lock-down restrictions due to COVID-19, I have not needed to travel, therefore I am rarely away from a power source. However, in my testing at home, I would compare the MacBook Air with my iPad Pro, consistently achieving 10+ hours of use and up to 16 hours depending on the workload.

The MacBook Air achieves an average of up to 9 hours additional battery life than my old 12-inch MacBook and Razer Blade 15 Advanced, whilst up to 6 hours better than my 16-inch MacBook Pro. This is a phenomenal achievement, which makes me excited about the prospect of an Apple Silicon powered 16-inch MacBook Pro.

In summary, my usage of the MacBook Air has been similar to an iPad, only charging every few days vs. every night. This is even true when using applications that would usually drain battery life at an accelerated rate, such as Zoom video conferencing, etc.

Compatibility

As part of the transition to Apple Silicon, Apple introduced Rosetta 2, which like the PowerPC to Intel transition, aims to allow software designed for Intel processors to run on the Apple M1. The image below shows Activity Monitor, which includes a new “Architecture” column highlighting the status of each application.

With memories of Rosetta, this is where I was most sceptical. I was fairly confident Apple would nail the performance and power efficiency, but compatibility can be a real challenge. For example, Microsoft has attempted (multiple times) to deliver x86 compatibility on Windows for ARM but continues to fail miserably.

The table below highlights which of my commonly used applications are currently running native Apple Silicon on or via Rosetta 2. Where available, I have also added the timeline in which a native version is expected.

It is important to note that this story is evolving daily, as developers update their applications. To stay up to date with the latest information, I would recommend reviewing the website “Is Apple Silicon Ready”. One of the strengths of the Apple ecosystem is the speed in which developers accept and adapt to changes, which appears to be the case with the Apple M1.

The big test for me was Microsoft Office, which I use regularly, with large, complex files. I also remember how badly Microsoft Office 2004 ran on Rosetta during the transition to Intel, therefore I was nervous to go back to this position.

Thankfully, as highlighted in the video below from Created Labs, Microsoft Office works remarkably well via Rosetta 2.

The same can be said for all of my common applications running via Rosetta 2. I occasionally see some slow-down during start-up, but once running, everything appears to work as designed, delivering acceptable performance and battery life. As an example, I re-ran Geekbench on the MacBook Air under Rosetta 2, which achieved the following result.

The single-core score reduced by 26% and the multi-core score reduced by 21%. This is a very impressive result, considering the M1 is still competitive against the Intel i9-9980HK in my 16-inch MacBook Pro.

Although I cannot prove this to be true, I am guessing Apple has worked some “hardware” magic to achieve this outcome. I find it difficult to believe that software alone could deliver this level of performance through emulation. Assuming accurate, this would highlight one of the benefits of transitioning to custom silicon, something that would not have been possible when transitioning from PowerPC to Intel.

It is worth noting that Rosetta 2 cannot translate kernel extensions or applications that rely on client virtualisation. Developers must also be aware that Rosetta 2 cannot translate AVX, AVX2, and AVX512 vector instructions, therefore some applications will simply fail to start.

A good example is Docker, which relies on Go for Docker Desktop and client virtualisation to run Linux containers. I use Docker frequently as part of my developer workflow, therefore was unable to use the Apple MacBook Air for development at this time. Thankfully, Docker is actively working on a new version, as highlighted in their recent blog post.

I was pleasantly surprised that Terminal tools can be used under Rosetta 2, such as the popular Homebrew Package Manager. To achieve this outcome, the “arch” command must be used, which allows the user to specify the architecture of a universal binary.

For example, the following command installs Homebrew under x86_64 emulation.

    arch -x86_64 /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/in...)"

Although functional, package compatibility can be inconsistent, therefore I hope to see native support soon. I would recommend keeping an eye on the Homebrew GitHub Issues Page for the latest updates.

Finally, client virtualisation, such as VMWare Fusion Player and Parallels Desktop are also unusable at this time. In macOS 11 Big Sur, Apple has provided access to a new virtualisation framework, which provides high-level APIs for creating and managing virtual machines. Therefore, I expect to see the availability of community projects soon, followed by the release of commercial applications from companies such as Parallels, etc.

Although some of these current restrictions impact my workflow, I do not expect them to affect mainstream consumers (those buying a MacBook Air). I assume by the time Apple release a true “Pro” Mac running Apple Silicon, these limitations will be resolved.

In short, with some known exceptions, Apple has nailed compatibility!

Apple tend to use the word “magic” to describe their products, ironically, Rosetta 2 feels like magic, but will likely be relegated to a footnote in the history of macOS. Therefore I commend the Rosetta 2 development team for their impressive implementation.

Conclusion

In my framing, I highlighted two possible outcomes regarding the transition to Apple Silicon.

The first described the impact if Apple were to succeed, by achieving better performance and power efficiency, delivered in a smaller form-factor at a lower cost.

Although it is still early days, my initial reaction is that Apple is on the right path!

In my experience, Apple Silicon is something special, delivering performance and power-efficiency that punches well above its weight, especially when you acknowledge that the MacBook Air I have been testing is passively cooled (no fans).

Thanks to Rosetta 2, Apple has also delivered a viable transition runway that makes selecting a Mac running Apple Silicon seamless for mainstream consumers. This achievement should not be underestimated, considering the past failures of Windows on ARM.

Assuming Apple can sustain the previously stated rate of innovation, as well as scale their silicon to “pro” devices, I am very excited by the future. The idea of an Apple M1x with more cores, more memory and higher clock rates is a tantalising proposition, something that I hope to see in early 2021 as part of the anticipated 16-inch MacBook Pro refresh.

As to where this leaves Intel/AMD and the x86 architecture, as a PC manufacture (e.g. HP, Dell, Lenovo), I would be very nervous. Although I do not anticipate any immediate impact, the mid/long-term success of these companies could rely upon their ability to partner, as well as remove existing barriers that have prevented innovation.

As a wild card, I will also be watching Nvidia very closely, who have demonstrated strong, sustained innovation across their graphics and AI technologies. With their recent announcement to acquire ARM, they potentially have all of the ingredients to rival Apple, especially if they form a close partnership with Microsoft and/or (finally) embrace Linux from a software perspective.

My only concern as a byproduct of Apple Silicon is that Apple becomes even more confident, further locking-down their ecosystem, making open hardware and software increasingly difficult to embrace. As a long-time PC enthusiast and open evangelist, this trend would be disappointing.

With that said, I cannot help but be inspired by Apple’s continued ambition and ability to execute! Many felt that the company would falter without Steve Jobs, however, I feel Apple may be on the verge of delivering another market-disrupting innovation!

Docker on Fedora 33

Tue, 10 Nov 2020 00:00:00 -0600

In the article “Linux at Work”, I documented my testing of a Dell XPS 13 running Fedora 33 as my primary device for work.

Thanks to the growth of web applications and frameworks such as Electron, the experience was mostly positive, with only a few limitations, caused by the lack of a native Linux Microsoft Office 365 applications (Word, PowerPoint, Excel).

The only other challenge I experienced was related to software repositories that had not yet been updated to support Fedora 33 (which has now been officially released). For example, Docker, which I use for software development to package and run applications in containers.

Thankfully, with support from the open-source community, it was relatively easy to get a Docker compatible environment running on Fedora 33, taking advantage of components offered by the Moby Project.

The article aims to document the installation process of a Docker compatible environment on Fedora 33.

Moby Project

Moby is an open-source project, created by Docker, to enable and accelerate software containerisation.

It includes a framework and library of components that provide the building blocks to set up a custom container-based architecture, primarily targetting container enthusiasts and professionals. The image below highlights the core components of Moby.

I would still recommend using Docker Community Edition (CE) for mission-critical development, but the Moby Project can be used to enable a Docker compatible environment if the official Docker channel has not been updated (often the case with new operating system releases).

Control Groups (cgroups)

As part of the Linux kernel, control groups (AKA cgroups) are a collection of processes that are bound by the same criteria and associated with a set of parameters or limits. For example, cgroups support the following capabilities:

Resource limiting: Limit memory usage, including the file system cache.
Prioritisation: Prioritise processor utilisation and/or disk I/O throughput.
Accounting: Measure resource usage, which can be used for billing purposes.
Control: Freezing processes, their checkpointing and restarting.

There are two versions of cgroups (cgroups v1 and cgroups v2) currently available. Starting with Fedora 31, cgroups v2 was enabled by default. Although this was a positive step forward from a Linux kernel perspective (and I respect Fedora for leading the way), unfortunately, Docker, Mobyand Kubernetes are not yet compatible with cgroups v2.

As a result, to enable Docker or Moby on Fedora 33, cgroups v1 must be re-enabled.

Install Docker on Fedora 33

The following three steps outline how to install a Docker compatible environment on Fedora 33.

Step One: The following Terminal command will re-enable cgroups v1 on Fedora.

sudo grubby --update-kernel=ALL --args="systemd.unified_cgroup_hierarchy=0"

Step Two: The following Terminal commands will ensure Docker has the required network access to manage containers.

sudo firewall-cmd --permanent --zone=trusted --add-interface=docker0

sudo firewall-cmd --permanent --zone=FedoraWorkstation --add-masquerade

Step Three: The following Terminal commands will install and enable Docker compatible components.

sudo dnf install moby-engine docker-compose

sudo systemctl enable docker

Once the installation is complete, reboot Fedora.

To test the installation, simply run the following Terminal command to run the “Hello World” container.

sudo docker run hello-world

The screenshot below highlights a successful installation.

That’s it! You should now have a Docker compatible environment running on Fedora 33.

Conclusion

As previously stated, I would still recommend Docker CE for any mission-critical development. However, if you are impatient (or simply inquisitive), the Moby Project is viable and worth a look.

Windows 10 ISO

Fri, 06 Nov 2020 00:00:00 -0600

In 2017, Microsoft introduced the terminology “Windows as a Service”, which many assumed would be a subscription model for Windows (like Office 365). Instead, it signified a change in approach regarding how Microsoft builds, deploys, and services Windows, shifting to regular “feature updates” that contain new features and fixes.

On paper, this approach sounds great, similar to how web services are continuously updated. However, Windows 10 is a complex beast, which makes testing very difficult, especially when executed under pressure to meet an arbitrary timeline.

As a result, there can be good reasons to avoid the latest release of Windows 10, especially for businesses, where consistency and reliability are generally prioritised above new features.

Unfortunately, due to the “Windows as a Service” approach, Microsoft does not make it easy to download older versions of Windows 10, always pushing you towards the latest and “greatest”.

Thankfully, the user “AveYo” has posted a useful “.bat” file on GitHub, which makes the process trivial.

As a prerequisite, download the official “Media Creation Tool” from Microsoft.

Once complete, download the “.bat” file from GitHub and place it in the same folder as Media Creation Tool.

Finally, open the Command Prompt (cmd.exe), navigate to the folder location and run the following command:

MediaCreationTool.bat

This will trigger a window to appear, with options to select any officially released version of Windows 10 (starting with 1507).

Once selected, the standard Microsoft Media Creation Tool is triggered, allowing you to create an ISO.

That’s it! You can now easily download and create installation media for any version of Windows 10.

Zero Trust

Sat, 31 Oct 2020 00:00:00 -0500

Historically, enterprise businesses have relied upon a traditional “Moat/Castle” strategy for IT Security, which aims to protect critical business assets through perimeter security controls (e.g. Firewalls, etc.) This strategy was established during a time when the majority of IT services existed within private data centres, where the end-to-end technology stack was fully managed.

Users would be “verified” at the perimeter before being granted access to business assets, usually via the Local Area Network (LAN) or remotely via a Virtual Private Network (VPN). Once inside the permitter, the business would assume a level of trust.

Unfortunately, modern working has exposed the Most/Castle strategy as inadequate, highlighted by three realities:

The majority of users (employees/contractors) no longer operate from controlled locations (e.g. Offices), especially post-COVID, where remote working has become a necessity.
IT Services are increasingly distributed (e.g. SaaS, Cloud Hosting) and rented vs. owned (e.g. Public Cloud, Colocation Data Centres), meaning business-critical assets could be hosted and/or processed at multiple unmanaged locations, making the concept of a permitter very difficult to define and maintain.
It is no longer safe to assume that verified users (internal or external) are “friendly”, recognising the growth in outsourcing, partnerships and social engineering.

Many enterprise businesses (and security vendors) continue to perpetuate the inadequate Moat/Castle strategy, often positioning additional security layers that deliver the illusion of increased control. Unfortunately, in the context of security, investment against the wrong strategy can be counterintuitive, adding complexity and technical debt, which contributes to additional risk.

With this in mind, a fundamentally different strategy is required, one that accepts and embraces the realities of modern working.

Zero Trust

Zero Trust is an emerging strategy for IT security, which assumes that internal and external threats exist at all times and that all networks are inherently hostile.

The traditional concept of a perimeter is disregarded, shifting the emphasis to Identity Access Management (IAM). It could be argued that identity becomes the new perimeter, stipulating that all connections (users, services, machines or devices) accessing business assets must first be authenticated and authorised.

In this scenario, IT services (applications and data) are individually secured and monitored at source via a central control pane, following the Principle of Least Privilege (PoLP). Therefore, a holistic security posture can be maintained regardless of the user or IT service location, with appropriate threat identification and controls that inherently limit the impact of a breach.

It should be noted, that although not depicted in the diagram, the firewall still plays an important role, shifting from Layer 3 (Network) to Layer 7 (Application), creating a Software-Defined Perimeter that can be managed from a central control pane.

The philosophy of Zero Trust is reasonably easy to understand (everything is hostile), but the practical implementation of a Zero Trust architecture can be challenging.

The following eight principles highlight the foundational technologies and techniques that enable a Zero Trust architecture:

Architecture and Asset Transparency: Situational Awareness is an important part of any strategy, therefore it is critical to understand the end-to-end architecture and assets across the business, covering the users and their devices, through to the services and data they are accessing. Without this basic insight, it becomes very difficult to identify the risks and dependencies that allow for data-driven decisions. In the context of Zero Trust, focus on assets that interact with a network (specifically important within a microservice architecture), leveraging automated discovery tools to reduce the burden of ongoing maintenance.
Prioritise Identity Access Management: In a Zero Trust architecture, identity becomes the new perimeter, it is important to have a single source of identity for all users, services, machines and devices, ideally managed via a central control pane and complimented by Privileged Access Management (PAM). These identities must be continuously maintained, modified and monitored throughout their access lifecycle, requiring a robust governance model, mature processes and discipline (e.g. Joiners-Movers-Leavers, etc.)
Authenticate Everything: Every network flow from a user, service, machine or device must be authenticated, preferably leveraging a modern authentication protocol (e.g. SAML, Oauth2, OpenID Connect, etc.) User authentication must include Multi-Factor Authentication (MFA), alongside the use of Secure Cryptoprocessor technologies (e.g. HSM, TPM, etc.)
Authorise Everything: The Principle of Least Privilege (PoLP) must be established, where all service and data access requests are only granted if required (need-to-know basis). Ideally, every request to access a service or data must be checked by a central policy engine, which should be dynamic and calculated from as many sources of data as possible, enabling conditional access.
Log Everything: Given that users, services, machines and devices are increasingly network connected, it is important that comprehensive logging and monitoring is maintained to provide visibility. The output can be used to help identify gaps and opportunities, as well as verify that all policies are being enforced appropriately. Any identified issue can be automatically flagged and remediated following an event-driven architecture.
Hostile Network: Recognising that all networks are considered hostile, all users, services, machines and devices must be secured at source and in transit. Network locality is not sufficient for defining trust, however, network segmentation concepts to create secure enclaves which limit network and application flows between workloads can form effective mitigation when looking to reduce the “blast radius” of a breach.
Health Confidence: The health of services, machines and devices are important signals that should be used as part of the conditional access policies. Health should be established based on standard control policies, which stipulate the minimum required criteria for access (e.g. hardening controls, patching levels, etc.) NCSC or CIS standards/benchmarks can be utilised to provide a robust foundation.
Automation: IT Security is a dynamically evolving and often volatile space, with new exploits and threat actors being identified regularly. With this in mind, it is unrealistic to rely upon reactive, manual interventions to protect an enterprise ecosystem. Investment in automation across the ecosystem can mitigate this challenge, promoting agility, whilst ensuring the consistent adoption of security controls that can be proactively and immediately applied based on specific criteria or events.

The diagram below helps to highlight the key components of Zero Trust:

Alongside these principles, a robust education curriculum must be established to help embed the concepts of Zero Trust. This is especially important for any team designing/delivering digital solutions (e.g. Product Teams), as well as architects and engineers working on the core infrastructure, DevOps or DataOps.

Zero Trust Anti-Patterns

Another perspective that can help position Zero Trust is to review anti-patterns, which include common designs that reduce the effectiveness of Zero Trust, resulting in technical debt. Zero Trust can only be effective if the entire business is aligned, although as with any transformation, the transition will take time.

Virtual Private Network (VPN): Client VPN and Business-to-Business VPN (B2B VPN) technologies used to establish connectivity between two parties (user or business). These technologies commonly operate at the Network Layer (Layer 3), maintained by brittle access control lists. VPN technologies often assume IT services are static and therefore are not well suited for highly ephemeral, dynamic services. As a result, without application layer controls, achieving the Principle of Least Privilege (PoLP) across multiple IT services via a single VPN tunnel can become very complex. Zero Trust alternatives would include targeting API-Centric connectivity (e.g. REST) and/or a service such as Boundary from HashiCorp.
Application Architecture: Traditional three-tier applications running within a private data centre are often tightly-coupled, leveraging proprietary technologies and custom integrations. This “monolithic” approach to application architecture commonly relies upon network locality to define trust, instead of targeting modern authentication protocols (e.g. OAuth2), open APIs (e.g. REST) and secure transit (e.g. TLS). To achieve Zero Trust, a modern “cloud-native” application methodology should be followed, such as The Twelve-Factor App.
IP Whitelists: Legacy services and data stores can rely upon an IPv4 address to define trust. Static IP addresses are increasingly rare, due to address translation technologies, proxies and dynamic services. They can also be subject to “man-in-the-middle” attacks and address spoofing. A Zero Trust alternative would be to implement a modern authentication protocol, such as OAuth2, OpenID Connect, etc.

Finally, like all trending “buzzwords”, vendors will look to “rebrand” their existing products and services to promote Zero Trust concepts (the same can be said for Artificial intelligence). Therefore, be careful to ensure that any investment is verified as “Zero Trust” compliant, by benchmarking it against the previously highlighted principles.

Conclusion

In conclusion, I am a strong believer that enterprise businesses need to stop perpetuating inadequate security strategies, recognising the hard reality that modern working requires a different approach.

Zero Trust is a good example of a modern security model, but the transition can seem insurmountable for an enterprise business with a large legacy ecosystem. Even in this scenario, it is critical to “stop the bleed” and look for pragmatic opportunities to invest in technologies and techniques that support the principles of Zero Trust, for example, I would recommend protecting (growing) investments related to Identity Access Management (IAM).

Nvidia Jetson Nano

Sun, 18 Oct 2020 00:00:00 -0500

In the article “Computer Vision”, I outlined my intent to build a small Computer Vision test lab, designed to support a range of business use cases. The article highlighted the target uses cases (internal and external), requirements, proposed architecture and bill of materials.

This article will detail the setup of the Computer Vision test lab, with a follow-up article that will focus on the testing observations and results.

The test lab is built around the Nvidia Jetson Nano Developer Kit, which is a small and power-efficient computer, with enough horsepower to run multiple neural networks in parallel, supporting applications such as image classification, object detection, segmentation, etc.

The full bill of materials can be found below, selected based on their interoperability.

Nvidia Jetson Nano Developer Kit: £101.50
SanDisk Ultra 64GB A2 microSD Card: £9.69
Geekworm 5V 4A 4000mA AC-DC Adaptor: £12.99
Raspberry Pi Camera v2.0+: £24.00
EDiMAX EW-7611ULB USB Wi-Fi Adapter with Bluetooth: £12.17

The total one-time cost of the test lab was £160.35, making it a cost-effective entry point for anyone looking to explore Artificial Intelligence or subfields such as Computer Vision.

Hardware Setup

The Jetson Nano Developer Kit is shipped in a small, bare-bones box, which includes the Jetson Nano module and carrier board. No other peripherals are provided (e.g. SD Card, Power Adaptor, etc.) The box does double as a fold-out paper stand, which is a nice touch, but fairly useless outside of the initial setup.

The back of the box highlights the product specification and board schematic, as well as acknowledging the need for an SD CArd and 5V 2A power source.

Although the Jetson Nano Developer Kit can be powered via micro-USB, I would highly recommend an external power adaptor to ensure a consistent and reliable 5V 2A. I selected the Geekworm 5V 4A 4000mA AC-DC Adaptor, which appears to work well (even under heavy load). To enable the external power adaptor, a jumper must be used to close the DC EN connection (highlighted in the photo below).

Although delivered as one unit, the Jetson Nano Developer Kit is two components, the Jetson Nano module, mounted on a carrier board. The module includes a large heatsink, with an option to add an external 40mm fan (recommended with hot or enclosed environments). The carrier board includes a range of input/output options, specifically HDMI, Display Port, 4x USB, 2x MIPI-CSI Camera Connector Connectors, 1x Expansion Header (GPIO, I2C, UART) and Ethernet.

The microSD slot can be found on the back of the Jetson Nano module, which can be accessed whilst mounted to the carrier board (highligted in the photo below).

I previously mentioned the fold-out paper stand, which works fine for the initial setup but does not include any mounting points for peripherals such as a camera, etc. As a result, I constructed a rudimentary stand out of Lego for the camera. The Raspberry Pi Camera connects via a MIPI-CSI Camera Connector.

To install the camera, simply disconnect all power sources (CCD cameras are sensitive to static) and connect the ribbon, ensuring the blue side is facing away from the Jeston Nano module (outwards).

Finally, the EDiMAX EW-7611ULB USB Wi-Fi Adapter with Bluetooth adaptor can be inserted into an available USB port.

This concludes the hardware setup of the Jetson Nano Developer Kit, ready to be connected to a monitor (HDMI or Display Port) and USB Keyboard/Mouse.

Software Setup

Nvidia provides a custom version of Ubuntu Linux, which can be downloaded from their website.

Once downloaded, follow the steps outlined in the “Jeston Nano Developer Kit Getting Started Guide” to write the operating system image to the microSD card. I used Windows to write the image, using an application called Etcher.

Once written, do not be surprised if the host operating system (in my case Windows) displays multiple errors when attempting to read the microSD card. Simply eject the microSD card and insert it into the Jetson Nano module.

The newly written operating system can now be used to boot the Jeston Nano Developer Kit, following the on-screen instructions to complete the initial operating system setup (e.g. Username, Password, etc.)

Once at the Linux desktop, it is time to set up and enable the EDiMAX EW-7611ULB USB wireless adapter.

Start by opening the terminal and cloning the following GitHub repository (you will temporarily need an Internet connection via the Ethernet port).

git clone https://github.com/lwfinger/rtl8723bu.git

Within the terminal, navigate to the newly downloaded directory and type the following command:

source dkms.conf

Type the following command to create a working project directory:

sudo mkdir /usr/src/$PACKAGE_NAME-$PACKAGE_VERSION

With the directory created, type the following command to move the files to the working project directory:

sudo cp -r core hal include os_dep platform dkms.conf Makefile rtl8723b_fw.bin /usr/src/$PACKAGE_NAME-$PACKAGE_VERSION

Add the files to DKMS with the following command:

sudo dkms add $PACKAGE_NAME/$PACKAGE_VERSION

Finally, the driver can be installed by typing the following command:

sudo dkms autoinstall $PACKAGE_NAME/$PACKAGE_VERSION

With the installation of the wireless adapter complete, restart the Jeston Nano Developer Kit, at which point wireless networks should be accessible via “Settings > Network”.

Although the Raspberry Pi Camera should be “plug and play”, it is worth testing. I would recommend using the CSI-Camera repository, which includes code that interacts with the MIPI-CSI Camera Connector. Using the provided Python script ‘python simple-camera-py’, the camera can be activated, with a live feed being displayed to the monitor.

Finally, the following two terminal commands will update the operating system and Nvidia Jetpack SDK, which includes a suite of tools to support Computer Vision use cases.

sudo apt-get update

sudo apt-get install nvidia-jetpack

That is it! The Jeston Nano Developer Kit is now ready to start processing Computer Vision workloads. In the next article, I will focus on testing observations and results.

Linux at Work

Thu, 08 Oct 2020 00:00:00 -0500

Over the past twenty-years, enthusiasts, researchers and analysts have predicted the “Year of the Linux Desktop”.

I started using Linux in 2001, during a strange period when boxed Linux distributions could be purchased in a physical store (e.g. PC World, Staples, etc.) This was short-lived phenonium, but I did personally buy a boxed version of Mandrake Linux (which at the time, included the best documentation).

Ironically, Linux has seen enormous growth in almost every market (e.g. Server, Mobile, IoT), but has failed to gain momentum on the desktop (currently estimated at 1.74% global market share).

With this in mind, I wanted to answer the question, could I use a Linux Desktop full-time for work?

As Chief Technology Officer (CTO), my time is split across the following tasks.

Collaboration (e.g. E-mail, Messaging, Conferencing, etc.)
Productivity (e.g. Document Creation, Reviews, Presentations, etc.)
Architecture (e.g. Creating and Reviewing High/Low-Level Designs, etc.)
Engineering (e.g. Software Development, Administrative, etc.)
Testing (e.g. Integration, System, Acceptance, etc.)

I would like to state that architecture and engineering are where I spend the majority of my time, but the reality of my role places a heavy emphasis on collaboration and productivity. I do however look for opportunities to stay grounded, by getting “hands-on keyboard” as often as possible.

Linux Work Setup

As a long-time advocate of Fedora from Red Hat, I selected the near-final version of Fedora 33 as my Linux distribution. Fedora 33 is set to be released in November and includes the usual updates to the Linux kernel (v5.8), GNOME (v3.38.1), as well as a controversial shift away from EXT to Btrfs as the default file system. As a long-time supporter of the ZFS file system, I have mixed opinions about this change, but happy to give it a try.

I installed Fedora 33 on a Dell XPS 13 (9300), with the following specification:

13-inch Dell XPS 13 (9300)
1.3GHz 4-Core Intel i7-1065G7 (Turbo up to 3.9GHz)
16GB 3733MHz LPDDR4x RAM
512GB M.2 PCIe NVMe SSD
Intel Iris Plus Graphics
13.4-inch Display (1920x1200, 500nits)

Although not a powerhouse, the Dell XPS 13 is a fast, portable laptop with the same exceptional aesthetics as the Dell XPS 17. The photo below compares the Dell XPS 13 against the 12-inch Apple MacBook, highlighting the similar dimensions, which is impressive considering the Dell is significantly more powerful.

Dell has also “confirmed” support for Linux (specifically Ubuntu), selling a specific “developer edition”.

The installation of Fedora 33 was flawless, with only the integrated fingerprint scanner failing to be recognised (I have not found a workaround yet). The lack of dedicated graphics within the XPS 13 is disappointing from a performance perspective but does remove a common challenge when installing Linux (especially when working with Nvidia graphics.

As I am currently working from home (due to COVID-19), the Dell XPS 13 is connected to my Samsung C49RG90 Super Ultrawide display and Bluetooth Logitech keyboard/mouse.

The 5120x1440 resolution of the display works the Intel Iris Plus Graphics fairly hard, but still provides enough horsepower to deliver a smooth experience for collaboration and productivity tasks. The only minor watch-out was the refresh rate, which could only be set to a maximum 100Hz (instead of the default 120Hz).

Linux Work Software

To support my work needs, I installed the following software, the majority of which is Free and Open-Source (FOSS).

The software installation process was fairly simple, following the steps outlined in my Fedora 31 article, using a combination of different package managers (DNF, Flatpak and Snappy). The only issue I experienced was with the SimpleNote snap package, which I resolved by using the traditional DNF install.

The overall software story on Linux is generally positive, with key applications such as Chrome, Zoom and 1Password all being available and offering a comparable feature-set to their Windows/macOS counterparts.

It is also great to see companies continuing to invest in Linux, for example, 1Password recently announced a native Linux client is in development, replacing the need to use a Chrome plugin.

At work, I am in the fortunate position where the majority of our applications are accessible via the browser (e.g. Corporate, Financial, HR, etc.) Therefore, assuming the user has a valid identity, multi-factor authentication enabled and meets the conditional access requirements, the underlying operating system is largely irrelevant (e.g. Windows, macOS, Linux, etc.) In this scenario, the browser becomes the operating system, placing a heavier emphasis on projects such as Chromium.

One area of contention is Microsoft Office 365, used for Collaboration and Productivity.

Over the past few years, Microsoft has demonstrated a renewed energy for Linux, with the release of Visual Studio Code and Teams (helped by the fact that they are both Electron applications). They have also announced the release of Edge (Chromium-based browser), which could have integration benefits for Microsoft 365 customers.

Unfortunately, the main Microsoft Office Suite (Outlook, Word, PowerPoint and Excel) are not natively available on Linux, forcing the use of virtual environments, emulation or Office Online.

Office Online has come a long way in recent years and I prefer Microsoft Outlook on the web over the desktop version. It is feature-complete (at least for my usage) and includes a “less cluttered” user interface, which streamlines common tasks.

Unfortunately, I cannot recommend the online versions of Microsoft Word, PowerPoint and Excel. Although they all technically work, I find the majority of my documents (which tend to be quite large and complex) result in constant stuttering and loading delays, as well as the occasional formatting inaccuracy (mainly with colours). These barriers make the user experience very cumbersome, resulting in constant frustration.

As an alternative, I tested LibreOffice (v7.0.x), WPS Office (v11.1.x) and OnlyOffice, which all include native Linux applications and compatibility with the Microsoft Office Open XML format (e.g. docx, pptx, xlsx, etc.)

Unfortunately, I did not find a perfect solution. In my testing, LibreOffice had the most comprehensive feature-set, but struggled with text styling and compressed images from Microsoft Office. The screenshot below highlights the issues (red box).

I found WPS Office to be more compatible, but less comprehensive. The screenshot below highlights the same slide, achieving a (slightly) better outcome.

I was most disappointed by OnlyOffice, which I had heard good things about (especially regarding Microsoft Office compatibility), but the applications felt very unoptimised on the desktop, delivering an undesirable user experience. Although I have not completed thorough research, I assume OnlyOffice for Linux is a shell for their web applications (making it only marginally better than Microsoft Office Online).

If I had to select one office suite for Linux, it would be WPS Office, which although not perfect, was the most compatible and delivered enough features to support my common use cases. The only watch-out is that WPS Office is not open-source (maybe a deal-breaker for some) and the license agreement requires a paid edition to support business use cases, starting at £31.00 per year (or £62 one time).

With that said, thanks to the open-source nature of Linux, I would likely recommend installing both LibreOffice and WPS Office, providing the best of both, depending on the specific task.

It is worth noting that to achieve font compatibility, you might need to install the Microsoft True Type fonts. Simply run the following Terminal commands:

sudo su

sudo dnf install curl cabextract xorg-x11-font-utils fontconfig

sudo rpm -i https://downloads.sourceforge.net/project/mscorefonts2/rpms/msttcore-fonts-installer-2.6-1.noarch.rpm

Once complete, the Microsoft fonts should be accessible across the operating system and compatible applications.

Finally, if all else fails, you can always run Windows in a Virtual Machine. Fedora comes pre-installed with Gnome Boxes, which provides a simple GUI for creating virtual machines.

I use VMware Workstation Pro, which includes a more comprehensive feature-set (3D Graphics, Advanced Networking), as well as compatibility with other VMware-based virtualisation technologies (vSphere integration, etc.)

VMware Workstation Pro requires a paid license but includes a 30-day free trial. VMware Workstation Player is also an option, which is a free but includes feature limitations (single virtual machine).

To install, simply download VMware Workstation Pro and run the following Terminal commands (update the filename as required):

sudo chmod 755 VMware-Workstation-Full-abc.x86_64.bundle

sudo ./VMware-Workstation-Full-abc.x86_64.bundle

Once complete, Vmware Workstation Pro can be used to install Windows 10 using the ISO.

Conclusion

Overall, when considering my role and common tasks, Linux is absolutely a viable operating system for work.

The primary pain point is Microsoft Office compatibility, but this is “workable” through the use of Linux compatible applications (LibreOffice and WPS Office) alongside Office Online. With that said, I would love to see Microsoft deliver a native Linux version of Office in the future.

Regarding the Dell XPS 13 and Fedora 33, both worked flawlessly. Recognising that Fedora 33 is still pre-release (scheduled for GA on 28-OCT-2020), some software repositories had not been updated (e.g. Docker), but this did not cause me any significant issues.

In conclusion, it may not be the “Year of the Linux Desktop”, but for me at least, the Linux Desktop is now a viable option as my daily driver.

Console Wars 2020

Sat, 03 Oct 2020 00:00:00 -0500

In 1995, at the age of eleven, I received a Sony PlayStation for Christmas (replacing my Sega Mega Drive). To this day, the iconic startup sound from the original PlayStation can’t help bring a smile to my face.

Over twenty-five years later, I have remained a PlayStation customer, owning seven consoles.

PlayStation (1995)
PlayStation 2 (2000)
PlayStation Portable (2004)
PlayStation 3 (2007)
PlayStation Vita (2012)
PlayStation 4 (2013)
PlayStation 4 Pro (2016)

Many of these consoles were pre-ordered ahead of the official UK release date and in the case of the PlayStation Portable (PSP), I went to the extra effort of arranging an import from Japan, nearly a year ahead of the UK launch.

The commemorative video released by PlayStation Access in 2013 does a nice job of highlighting the legacy of PlayStation (applicable to UK audience).

Alongside PlayStation, I have always owned a gaming PC, providing access to content that was not available on the PlayStation or benefited from additional graphics performance (e.g. The Elder Scrolls, etc.) I have also owned a Microsoft Xbox 360 and Nintendo Switch, although the Xbox only received modest usage.

In November, both Sony and Microsoft will release their “next-generation” consoles, specifically the Sony Playstation 5 and Microsoft Xbox Series X and Xbox Series S.

Based on my history, it would be safe to assume that I would target a PlayStation 5. However, I believe this generation signifies a turning point in the “console wars”, potentially changing the future of gaming (and my purchasing decision).

PlayStation vs. Xbox

On the surface, this impending console generation feels like any other, with two “corporate giants” going head-to-head with new consoles.

Sony has lent into the design language they first introduced with PlayStation VR, delivering a futuristic two-tone design.

Microsoft has gone in a different direction, targeting a simple, minimalistic design for the Xbox Series X and Xbox Series S.

As to which console looks the “best” is completely subjective. I like elements of both consoles but have been a little surprised by the size of the PlayStation 5, which at 39cm tall, is an imposing piece of hardware. With that said, the size of the console does not impact me directly, as all of my AV equipment is hidden from view in a cupboard.

Hardware Specification

In many respects, the look of the console is irrelevant. The real story is the hardware specification, which provides our first insight into the types of gaming experience we can expect from the next generation.

All three consoles leverage the Zen 2 and RDNA 2 architecture from AMD, delivering significantly better performance per wat when compared with the previous console generation (making a direct comparison difficult). The new architecture also delivers additional capabilities for developers to exploit, for example, hardware-enabled Ray Tracing (referenced in my previous article focused on the GeForce RTX series).

With this in mind, all three consoles should have similar capabilities, simplifying cross-platform development. However, how the architecture has been implemented can have a significant impact on the overall experience.

On paper, it would appear that the Xbox Series X has the performance edge as it relates to Processor and Graphics, while the PlayStation 5 takes the lead regarding Storage performance. The real-world impact of these decisions will likely not be known until later in the generation, once developers have learnt how to maximise the potential.

If I were to guess, I would expect cross-platform games to perform better on the Xbox Series X, with only the first-party games from Sony taking advantage of the additional storage read/write performance. This is already evident in some of the pre-release game trailers, such as Ratchet and Clank.

An area that I feel could be a challenge for both consoles is storage capacity. Although specific information has not yet been revealed, it is safe to assume a minimum of 150GB will be reserved for the formatting overhead and operating system, resulting in the following “usable” space.

Sony Playstation 5 = 675GB (Six 100GB Games)
Microsoft Xbox Series X = 874GB (Eight 100GB Games)
Microsoft Xbox Series S = 362GB (Three 100GB Games)

Knowing that many “Triple-A” games from the previous generation required a 100GB+ install, managing storage could become a continuous battle, especially for the consoles without an optical media drive.

Sony and Microsoft may deliver new compression techniques and/or provide an ability to delete unused game content (e.g. game modes, textures, languages), but these techniques will only soften the impact of the issue, not resolve it.

Thankfully, both Sony and Microsoft have confirmed that backwards compatible games can run from USB storage (e.g. External SSD), which should still offer performance benefits thanks to the high-speed USB interface (up to 10Gb/s on Playstation 5) and the improved decompression performance of the Zen 2 architecture.

Xbox Series S

The Xbox Series S is a unique proposition, delivering a compelling price point (£249.99) and the promise of “next-generation capabilities” at a lower resolution (likely 1080p or 1440p).

Looking at the specification, this feels viable, with the same core architecture (Zen 2 and RDNA 2), but rightsized to hit a specific performance target.

Unfortunately, I can foresee two issues:

The reduction in memory (10GB from 16GB) and significantly lower bandwidth (224GB/s from 560GB/s) will likely become a difficult bottleneck for developers to manage.
As previously highlighted, the included 512GB storage (approximately 362GB usable) will be very limiting, especially without an optical media drive. Therefore, users will likely need to invest in expandable storage. At £219.99 for a 1TB storage expansion card, the Xbox Series S would become more expensive than the Xbox Series X (which comes with 1TB as standard).

In short, the Xbox Series S feels like a high-risk purchase at this time and I would recommend seeing how well it is supported by developers before making a purchase.

Gaming as a Service

Success for the previous console generations has been driven by the number of consoles sold. In a world where requiring the console was a prerequisite to play games, this success criteria made perfect sense.

However, like other entertainment industries (e.g. television, music, film), gaming is evolving and I believe Microsoft has anticipated the shift.

With the launch of Xbox Game Pass and Xbox Cloud Gaming (game streaming), Microsoft has begun to reset what it means to be a gamer, delivering a service that is accessible across multiple platforms (e.g. PC, Console, Mobile, etc.) This approach removes the emphasis on the console itself, positioning it as just one part of a larger ecosystem.

With this in mind, I believe the new success criteria for gaming should be the number of customers actively playing games and therefore investing in services. In this new world, I feel Microsoft has three critical ingredients to achieve success.

Online Services: With the launch of Xbox Live in 2002, Microsoft has been the clear market leader in online services. Sony responded with the Playstation Network and Playstation Plus but has always been playing catch-up, with multiple significant setbacks, including the infamous PlayStation Network Outage. In areas where Sony have attempted to innovate, such as PlayStation Home, user adoption has been underwhelming. In short, Microsoft has demonstrated throughout their entire business an ability to successfully evolve from traditional sales models to online subscription services (e.g. Microsoft 365, etc.)
Cloud Infrastructure: As online services and cloud streaming becomes increasingly important, having a scalable, reliable, and globally available cloud infrastructure becomes increasingly important. This is an area where Microsoft can lean back on their other multi-billion dollar business, specifically Microsoft Azure. With over 160 physical data centres around the world, supporting the critical business operations of every major industry, Microsoft can leverage this sizable investment to cost-effectively innovate and scale their gaming capabilities. In comparison, Sony must rely upon third-party data centres (such as Equinix to scale out their cloud infrastructure, which although is a viable business model, is a sizable investment when targeting a single revenue stream. Another option for Sony would be to partner with another hyperscale cloud provider (e.g. Amazon Web Services or Google Cloud Platform), however, both Amazon (Luna) and Google (Stadia) have their own gaming ambitions, creating a potential conflict of interest.
Finacial Foundation: Over the past quarter, Microsoft generated $73 billion of net cash, compared to $17.6 billion at Sony. Over the past four quarters, Microsoft generated $45 billion in free cash flow compared to $9 billion at Sony. In short, Microsoft has very deep pockets, that are only likely to deepen, based on the increased demands of cloud, productivity and collaboration capabilities in a post-COVID world. Although money is not a guarantee for success, it does present Microsoft with flexibility, for example, the recent acquisition of ZeniMax Media for $7.5 billion, which is a holding company for Bethesda Game Studios, id Software, MachineGames, and Tango Gameworks. I suspect this will be first of multiple major acquisitions we will see from Microsoft this generation, as they look to strengthen their first-party developer portfolio.

Leveraging these three ingredients, Microsoft has positioned two gaming services that I feel signify a change in the industry.

Xbox Game Pass: In 2017, Microsoft introduced Xbox Game Pass, delivering “the Netflix of gaming”. In the UK, for £9.99 per month, the user gains access to Game Pass Ultimate, which includes a library of over 250+ games accessible across Xbox, PC and Mobile (via Xbox Cloud Gaming). The library includes “release day” access to all first-party Microsoft Studio games (e.g. Halo, Forza, Gears of War, Flight Simulator), alongside games from key partners, including all 70+ games covered by EA Play. With the acquisition of ZeniMax Media, games such as The Elder Scrolls, Fallout and Doom will also be available. Considering each one of these games can cost up to £70 alone, £9.99 per month for everything, across multiple platforms, with no release delays, is a phenomenal deal.
Xbox All Access: As an extension of Xbox Game Pass, Microsoft also offers “Xbox All Access”, which bundles the console itself alongside Xbox Game Pass Ultimate over a 24 months subscription. As an example, Xbox All Access with an Xbox Series X costs £28.99 per month, which is £695.76 over the 24 months subscription. If purchased separately, the Xbox Series X and Xbox Game Pass Ultimate would cost £713.75, making “Xbox All Access” a compelling deal for anyone looking for a cost-efficient entry point into the next generation.

The End of PlayStation?

Is this the end of the road for Sony PlayStation?

No, although I feel this generation will mark a turning point in the gaming industry, Sony has two major strength areas… Brand and Exclusives.

The PlayStation brand is loved around the world, which like Apple, can carry a lot of weight. Alongside the brand itself, Sony has continued to deliver some of the best gaming experiences from their first-party studios, which they generally make exclusive to PlayStation.

Games such as God of War, Uncharted, The Last of Us, Spiderman, Gran Turismo, Little Big Planet, Horizon Zero Dawn, Demon Souls are critically acclaimed and arguably among the best games of the previous generation, and I see no sign of this changing in the short-term.

Sony also has Playstation Now, which is a largely underrated service that is comparable to Xbox Game Pass Ultimate. The foundations of PlayStation Now are built from their acquisition of Gaikai, delivering game streaming to PlayStation consoles and the PC (no mobile at this time). PlayStation Now also offers the ability to download select PlayStation 4 games, similar to Xbox Game Pass Ultimate.

At £8.99 per month (or £49.99 per year), with over 700 games available for streaming (300 of which are downloadable), the service is price competitive.

The major challenge for Sony is that new games are not immediately available on PlayStation Now, with no guarantee (at this time) that they will ever transition to the service. Therefore, unlike Xbox Game Pass Ultimate, users will still need to pay up to £70 per game for the first-party exclusives. This is what makes Xbox Game Pass Ultimate unique, where Microsoft can take advantage of their previous highlighted financial foundation.

Conclusion

I predict this generation will be defined by gaming services, not console sales.

With services such as Xbox Game Pass, Xbox Cloud Gaming and Xbox All Access, I believe Microsoft is best positioned to capitalise on this market shift.

Assuming this prediction is accurate, I believe this generation will end the dominance of Sony PlayStation, applying additional pressure on Sony as a business.

With this in mind, am I recommending an Xbox Series X?

In my opinion, anyone looking to join the next generation of gaming, who does not have access to multiple games platforms (e.g. PC), should seriously consider Xbox Series X and Xbox Game Pass Ultimate, potentially via Xbox All Access. I believe (today) this combination offers the greatest “bang for your buck”.

However, if you have access to a gaming PC, you might want to consider PlayStation 5, giving you the best of both worlds. Microsoft has committed to the continued support of the PC for their first-party studios, therefore the PlayStation 5 could be the perfect compliment, delivering access to their excellent first-party exclusives. Although more comprehensive, this approach is clearly more expensive.

As for me, knowing that I have a gaming PC, I will likely pick-up a PlayStation 5 over an Xbox Series X.

Foldable Future

Fri, 02 Oct 2020 00:00:00 -0500

With the launch of the iPhone in 2007, Apple redefined what it meant to be a smartphone, resulting in a decade of “black mirror” devices that have evolved through a range of sizes but retained the same basic shape and function.

In 2019, Samsung revealed the first foldable smartphone, known as the Samsung Galaxy Fold, which attempted to break the mould, by delivering an innovative new form factor.

The idea was simple, produce a product that could operate as a traditional smartphone, but unfold to deliver a more immersive “tablet-like” experience.

Unfortunately, the product launch was a disaster, with some serious quality assurance issues related to the hinge that resulted in a product recall.

Although a complete failure, Samsung clearly felt the core concept was a winner, as they proceeded to release a follow-up in 2020, known as the Samsung Galaxy Z Fold2 5G.

I have been testing this product, not because I believe it is ready for mainstream consumers, but as a device that I hope will provide an insight into the future of foldable smartphones.

The Fold2 (I refuse to use the full name) is a true flagship, incorporating just about every modern smartphone feature, with the exception of water resistance. The full specification can be found below:

Samsung Galaxy Z Fold2 5G
Qualcomm SM8250 Snapdragon 865+ (Octa-Core)
6.2-inch HD+ AMOLED Display (2260x816 @ 60Hz)
7.6-inch QXGA+ HDR10+ AMOLED Display (2208x1768 @ 120Hz)
12GB RAM
256GB Storage (UFS 3.1)
2x 10MP Selfie Cameras
3x 12MP Cameras (Ultra Wide, Wide-Angle, Telephoto)
4500mAh Dual Battery
5G, LTE, Wi-Fi 6, Bluetooth v5.0
Finger-Print Reader and Face Unlock
USB-C and Wireless Charging

Unsurprisingly, this specification comes at a very premium price point, starting at £1799. This (rediculous) price is the primary reason not to buy the Fold2, recognising that competent smartphones such as the OnePlus Nord can be purchased for £349.

Instead, the Fold2 should be seen as a concept, which happens to be available for purchase for those that have the financial means and desire/need to test future capabilities.

This article will focus on my observations/experience using the Fold2, to understand the viability and practical benefits of a foldable smartphone.

As an example, science fiction has hypothesised this type of technology for many years, as highligted by the image below, which was taken from the popular TV series Westworld.

If you are looking for a traditional smartphone review, I would recommend watching the video below from Marques Brownlee.

Before focusing on my specific observations, I want to acknowledge that I believe the Fold2 is the “real deal”, delivering on the promise of a foldable product that can replace a traditional smartphone and a small tablet.

It would be easy to see the Fold2 as a niche product or a gimmick, but after using the device for several weeks, I am convinced that foldable smartphones deliver tangible value, even for mainstream consumers.

I am also confident that Samsung has successfully evolved the hinge, delivering a truly innovate solution that appears to mitigate the previous issues.

The video below describes how the new hinge and sweeper technologies work, which have apparently been tested to withstand 200,000 folds (100 folds per day for five years).

With that said, there are a few areas that need additional work (either product evolution or time to mature).

The 22.5:18 aspect ratio of front display on the Fold2 is not ideal, making text input and/or media playback awkward. However, it is a positive step from the first generation, which included a smaller 25:9 display. The display is also limited to 60Hz, which is comparable to the Apple iPhone 12 but noticeably less responsive than the unfolded 120Hz display.

In my opinion, if a future foldable smartphone could deliver a more traditional aspect ratio for the front display, it could successfully achieve “the best of both”, operating as a traditional smartphone for day-to-day tasks, as well as a compelling small tablet, similar to the Apple iPad Mini.

As an example, the photo below highlights two iPhone’s, each with a 5.8-inch display, which could be “unfolded” into a larger 7.5-inch display. For comparison, I have placed the iPhone’s on top of the Fold2, demonstrating that a subtle shift in aspect ratio could deliver a better overall experience.

Another minor point of contention is the crease left by the fold itself, which is noticeable via sight and touch when using the unfolded display. However, in my experience, similar to the “notch” on the iPhone, when immersed in content, the crease is rarely a distraction. With that said, I do hope the technology continues to mature, delivering a truly flat display when unfolded.

At 16.8mm the thickness of the Fold2 when folded can be awkward, especially when combined with the “tall” aspect ratio. It is also heavier than an average smartphone at 282g, although the weight was only a minor inconvenience.

Although not directly comparable, the Microsoft Surface Duo has demonstrated the potential for a thinner foldable device at 9.9mm.

In my opinion, somewhere between the two (12mm) should be viable (minimising trade-offs), delivering an acceptable balance when you consider the technology required to achieve the folding mechanism.

The primary limitation is the need to manage the curvature of the OLED display, meaning it cannot close fully end-to-end. This is a limitation of the current generation OLED displays, which will hopefully improve in the future.

Unsurprisingly, based on the specification, the performance of the Fold2 is excellent, comparable to other Android flagship smartphones such as the OnePlus 8 Pro, etc. Battery life has also been very good, easily lasting a day with a varied workload include streaming video.

The same can be said for the five (yes, five) cameras, which are all flagship quality, similar to what can be found on the Samsung Galaxy S20 line of smartphones. Assuming “under-display” cameras become viable soon, manufactures could also remove the need for the “punch-hole” cameras, delivering a fully interrupted display.

Application compatibility is good, but not perfect. Due to the lacklustre adoption of Android tablets, developers have less incentive to optimise their apps for larger displays. Therefore, not all apps utilise the additional display real-estate effectively, for example, apps such as Facebook, Instagram, etc.

With that said, the transition between the front and unfolded display is seamless, with most apps delivering a better (if not fully optimised) experience on the larger display. For example, productivity apps such as Microsoft Office 365 and video playback apps such as Netflix work great, demonstrating the potential of a larger display!

Starting with Android 10, Google has made foldable smartphones a core part of the Android ecosystem. Therefore, I expect compatibility to improve over time, hopefully, incentivising hardware manufacturers and third-party developers to take advantage of the new capabilities.

The option to enable “usable” multi-tasking is especially intriguing, which has been a long-running limitation of smartphones that can impact productivity for content creation.

Conclusion

I am convinced foldable (or expandable) smartphones are the future! The ability to seamlessly transform a “traditional” smartphone into a tablet is breathtaking and feels like something straight from science fiction.

Outside of the “wow” factor, the foldable display also offers tangible benefits, potentially removing the need for a user to purchase/carry a smartphone and a tablet.

As smartphones continue to become more performant and feature-rich, it is even possible that a foldable smartphone could become the “one device” for mainstream consumers, even replacing the need for a laptop (assuming compatibility with a monitor and peripherals, etc.)

If I were Google, I would see foldable smartphones as a real opportunity for Android to innovate and differentiate, whilst also reinvigorating the dwindling tablet market. Therefore, I hope to see Google continue to prioritise native foldable capabilities in future versions of Android, especially as I predict Apple will not be far behind.

Overall, based on the constructive feedback outlined in this article, I believe we are two generations (two-years) from achieving a “refined” foldable smartphone that is viable for mainstream consumers. The cost will likely remain high but hopefully normalise to compete with regular Android flagships, which can be purchased for around £1000.

With an additional two generations (total of four-years) the technology should be mature enough to drive economies of scale, resulting in mid-range foldable smartphones that open the door to the mass market.

I am excited by this future, which I believe will signify the next major evolution of the smartphone!

Digital Transformation

Wed, 30 Sep 2020 00:00:00 -0500

If you work in enterprise business, I suspect you have heard the words “Digital Transformation”, which is often linked to the corporate strategy.

As an IT organisation, the words Digital Transformation should be seen as an opportunity to re-position IT, placing a heavier emphasis on value creation through the design and delivery of innovative digital products, services and insights.

A common mistake when targeting Digital Transformation is to simply add technology to an existing traditional business model. This approach can certainly drive value through digital efficiency, but true Digital Transformation must rethink the business model itself, potentially resulting in market disruption.

Thankfully, as demonstrated by many external case studies, Digital Transformation does not need significant investment and/or resources to be successful. In my experience, successful Digital Transformation initiatives start small, with a clear outcome and focus, the required sponsorship and a ruthless desire to execute and learn, prior to scale.

I recently delivered a short presentation (embedded below) that describes an approach to Digital Transformation, encompassing concepts such as Outcome over Output and Products over Projects. Having the right approach (a digital operating model) is a critical prerequisite when looking to innovate and deliver customer-centric digital products.

The approach highlighted in the presentation encapsulates my thinking, influenced by many sources, specifically Chris Shinkle a Product Manager at SEP.

Although having the right approach is critical, other common barriers include:

The digital initiative failed to effectively communicate with the internal stakeholders the desired outcome, specifically the goal and purpose.
The digital initiative attempted to add technology to an existing business model, instead of rethinking the business model itself. This approach can drive digital efficiency, but not digital transformation.
The digital initiative lacked focus, discipline, perseverance and/or patience, often attempting to “boil the ocean” and/or achieve a positive ROI within unrealistic timelines (Steve Jobs quote: “if you really look closely, most overnight successes took a long time”).
The digital initiative failed to implement a digital operating model, specifically a focus on product management, iterative development and continuous feedback loop.
The digital initiative did not have a clear data strategy, where relevant primary data sources are identified and accessible, delivering actionable insights.

To overcome these barriers, I promote the following four Digital Transformation principles:

Make digital an embedded part of the business, instead of attempting to manage a separate internal corporate venture cycle and/or separate P&L organisation.
Outline a clear strategy (Objectives and Key Results), but start small, with a small team, controlled investment and a measurable value proposition.
Clearly define the data strategy, including how to gain access to the required data sources and produce the required insights.
Accept that digital transformation is a long-term investment that may not achieve a positive ROI immediately.

Therefore, prior to pitching or sponsoring any new digital initiative, I ensure these principles are fully realised.

Computer Vision

Fri, 25 Sep 2020 00:00:00 -0500

In the context of digital transformation, Artificial Intelligence (AI) presents a tremendous opportunity to disrupt traditional business models.

I am personally interested in Computer Vision (CV), which is categorised as a subfield of Artificial Intelligence (AI) and includes many interesting business value opportunities that could drive digital transformation.

A common misconception is that digital transformation initiatives require significant resources and investment. This is simply not the case, with most successful initiatives starting small, with a clear outcome and a ruthless focus on execution through iteration.

With this in mind, I plan to create a small Computer Vision test lab for education and testing purposes.

What is Computer Vision?

Computer Vision is a field of study that seeks to develop techniques to help computers “see” and understand the content of digital images such as photographs and videos.

In short, the goal of a Computer Vision problem is to use observed image data to infer something about the world.

As previously stated, Computer Vision is a multidisciplinary field, often categorised as a subfield of Artificial Intelligence (AI) and Machine Learning (ML), which may involve the use of specialised methods and general learning algorithms.

Common Computer Vision techniques include:

Image Classification: Classification based on contextual information in an image.
Object Detection: Define objects within images.
Object Tracking: Follow an object of interest in a given scene.
Semantic Segmentation: Understand the role of each pixel in an image.
Instance Segmentation: Segments different instances of classes.

The proposed Computer Vision test lab will provide a real-world environment to explore these techniques, looking to identify strengths, weaknesses, barriers, etc.

Use Cases

Computer Vision is an exciting field of study, which continues to gain momentum across many industries.

In the context of work, the list below highligts a few examples where Computer Vision could potentially add value.

Anomaly Alert System: Provide real-time surveillance of a specific location, autonomously alerting users when a specific object enters the scene and/or executes a pre-defined movement.
Security/Privacy Filter: Monitor live video streams (e.g. video conferences, webinars), automatically detecting and redacting specific objects (e.g. locations, individuals, displays).
Employee Social Distancing: As a response to the COVID-19 pandemic, help users practice safe social distancing (physical separation) by monitoring office locations, notifying users of any risk.
Fever Control: Monitor human temperature, detecting potential fever symptoms, delivering an immediate notification to the user.
Accessibility: Support users with a visual impairment to remain safe and productive within the workplace.

The proposed Computer Vision lab must be capable of testing these use cases.

Requirements

I have defined the following high-level requirements to support the creation of a Computer Vision test lab.

The Computer Vision Test Lab capital investment must be low (< £200.00), with minimal operating costs (< £100.00pm).
The Computer Vision Test Lab must be viable for use at rural locations, including limited (intermittent) network connectivity and power consumption (< 10 Watt).
The Computer Vision Test Lab must accurately represent a “real world” production architecture, including edge infrastructure (e.g. Camera, Compute, Storage).
The Computer Vision Test Lab must support modern AI/ML software libraries, including CUDA, cuDNN and TensorRT.
The Computer Vision Test Lab must leverage open-source licensed software libraries (e.g. OpenDataCam, etc.) Ideally targeting MIT, GPL 2.0/3.0 and Apache Licence 2.0.
The Computer Vision Test Lab must support autonomous operation, without physical or virtual human intervention.
The Computer Vision Test Lab must include appropriate software and physical security controls (Zero Trust), including the ability to physically secure all compute/storage devices.

These requirements aim to ensure the Computer Vision test lab has a defined scope and can be easily reproduced.

Architecture

The high-level diagram below outlines the proposed Computer Vision test lab architecture.

I have selected an Nvidia Jetson Nano Developer Kit as my edge compute/storage capability.

The Nvidia Jetson Nano is a small, reasonably powerful computer that can run a wide range of advanced networks, including the full native versions of popular ML frameworks such as TensorFlow, PyTorch, Caffe/Caffe2, Keras, MXNet, etc.

The table below highlights the performance benchmarks, which are commonly multiple times better than completing edge computing capabilities (e.g. Raspberry Pi, etc.)

When combined with the Nvidia JetPack SDK, the Nvidia Jetson Nano delivers an easy-to-use, cost-effective (£100) and power-efficient (5W) compute/storage capability, capable of supporting image classification, object detection, segmentation, and speech processing, etc.

Bill of Materials

The proposed Computer Vision test lab architecture is very cost-effective. I have selected the following components based on their interoperability.

Nvidia Jetson Nano Developer Kit: £101.50
SanDisk Ultra 64GB A2 microSD Card: £9.69
Geekworm 5V 4A 4000mA AC-DC Adaptor: £12.99
Raspberry Pi Camera v2.0+: £24.00
EDiMAX EW-7611ULB USB Wi-Fi Adapter with Bluetooth: £12.17
Microsoft Azure: £75.00pm (Variable)

The total one-time cost is £160.35, with a variable monthly cost from Microsoft Azure (estimated at £75.00).

Conclusion

Over the coming weeks, I plan to have the Computer Vision test lab up and running, at which point I will document the setup and initial findings.

Battery Health

Fri, 11 Sep 2020 00:00:00 -0500

Although I love my Razer Blade 15 Advanced, I fear battery degradation (longevity) will be an issue.

For example, thanks to the vrtical integration between hardware and software, Apple are known for achieving exellent battery life across their portable devices. However, what is less commonly known is that they also do a nice job of protecting the battery health, extending the usable life of the battery itself.

After one-year, I have already noticed a significant drop in battery life with my Razer Blade, Which is likely due to poor battery management during charging cycles, etc. This issue is not limited to Razer laptops and likely an issue with all Windows-based portable devices, recognising that the software can play an important role in battery health.

Windows (by default) does not provide much insight into battery health (only the current charge), but thankfully it is possible to get Windows 8/10 to generate a comprehensive battery report.

Simply open a command prompt (CMD) and run the following command.

powercfg /batteryreport

This command will create a HTML-based battery report, found in the following location.

C:\Windows\System32\battery-report.html

Although the battery report does not specificlly highlight “health”, it does provide a lot of information to provide an educated understanding. The key sections include:

System Information
Battery Information
Recent Usage
Battery Usage
Usage History
Battery Capacity History
Battery Life Estimates

The report can be quite large, but is formatted well, with clear sections.

Each section also includes a summary (conclusion), which can be used as a quick “health check”.

Looking at the battery report from my Razer Blade, I do not see the battery lasting much beyond two-years (with usable battery life). Thankfully, unlike a modern Mac, the battery is easy to replace, assuming a trusted alternative can be located, as Razer do not sell batteries seperately.

Kid's Coding

Fri, 04 Sep 2020 00:00:00 -0500

As a technologist and a father, I had been eagerly awaiting the day I could share my passion for technology with my children.

This journey began last year when my four-year-old son and I started to explore the world of software development, using an app called Swift Playgrounds from Apple. I documented our experience in the article “Swift Playgrounds”.

Recognising that my son was not yet able to read, I was impressed by how quickly he was able to understand and executed the “Learn to Code 1 (Fundamentals of Swift)” playground, covering basic programming concepts such as Commands, Functions, Loops, etc. The interactive and fun nature of Swift Playgrounds helped to make the learning process engaging (gamified), with enough visual cues that enabled my son to recognise key letters (camelCase), without needing to read the syntax.

At the time, we stopped at the “Learn to Code 2 (Beyond the Basics)” playground, which upon review, would likely be beyond the capabilities of a four-year-old (rated for ten and above). I did not want to lose momentum by replacing enjoyment with frustration, which would likely impact future engagement.

Encouraged by our initial success, I started to research other viable software development learning experiences that would be suitable for my (now five-year-old) son. I was not worried about the specific programming language, instead, targetting options that focus on core concepts, whilst hopefully complimenting the development of other key skills (e.g. maths, reading, writing, problem-solving, etc.)

With five months of homeschooling behind us (due to the COVID-19 lockdown), I had plenty of time to introduce and trial some different options.

This article will provide a brief overview of the options we explored, including my initial observations.

Sphero

Thanks to a hackathon that I sponsored in 2017, I happen to own a Sphero Mini, which includes a free integrated development environment.

As part of the Sphero Edu app, Sphero offers a declarative development (no code) capability, similar to Scratch. It supports a wide range of platforms (e.g. iOS, Android), making it very accessible, assuming you are comfortable paying £44.99 for the Sphero Mini.

The app works well (very easy to set up) and provides multiple options for almost any level of experience. For example, the Sphero Mini can be programmed by drawing a path, using a sequence of code blocks, or (when ready) via JavaScript.

My son enjoyed controlling the Sphero Mini via the path drawing and was able to get started without any assistance. We quickly progressed to using the code blocks, where he required guidance due to the heavier emphasis on syntax. To ensure we had a challenge, we created a simple course using Mega Blocks, which we then attempted to programme.

It is fair to state that the Sphero Mini is not overly accurate (not helped by the carpet), therefore it took several attempts to achieve the desired outcome.

In summary, the connection between the software and a physical object was a real winner! It helped bring the code to life, clearly demonstrating the consequences of any changes, which my son found hilarious as the Sphero Mini regularly bounced off the walls.

The use of the path drawing capability was quick and simple but did not provide much in the way of learning. The code blocks started to introduce core concepts (functions, etc.) However, was more challenging due to the reliance on written code.

With that said, I commend the Sphero team for providing a comprehensive development environment, with a very low barrier to entry.

Sony Aibo

With everyone currently working from home due to the COVID-19 pandemic, I am current “dog sitting” our office robot dog (Sony Aibo).

Like Sphero, Sony offer development tools for Aibo, which can be accessed via the browser. This includes API documentation, as well as a declarative development (no code) capability.

Recognising the similarities to the Spehro Edu app, my son and I connected the development environment to Aibo and created some simple routines (e.g. Getting Aibo to dance, bark, wee, etc.)

I found the browser-based development environment to be less refined, but certainly functional. My son utilised the touch-screen of my laptop, as he is less familiar with traditional input mechanisms (e.g. trackpad, etc.)

Similar to Sphero, my son and I were able to create multiple routines by dropping and dragging code blocks. We worked together, with me reading the available commands and him creating the simple functions. Again, the ability to see the code trigger real-world events was a particular highlight, although I do feel for Aibo, as my son quickly increased the complexity of the routines (resulting in a refusal to obey).

Although our approach was certainly viable, it is clear that Sony (like Apple and many others) are targetting an audience that can read (not a surprise). Therefore, my son would not be able to progress without supervision and guidance.

Considering the price of a Sony Aibo (£2300), I am certainly not recommending it as a personal learning tool (maybe if you work as a teacher in STEM), but it is worth noting that many “toys” (such as Lego) now include simple development capabilities.

Hopster Coding Safari

Next, we tried Hopster Coding Safari, which is a free app (with in-app purchases) on the Apple iOS App Store.

This app resembles the first few chapters of the “Learn to Code 1 (Fundamentals of Swift)” playground, where the user is presented with a series of logic puzzles to solve.

The app is well built, with good graphics and an intuitive user interface. The logic puzzles involve navigating a cartoon animal to a defined endpoint using drag and drop programming, presented as symbols, instead of written syntax.

My son (and his younger sister) were immediately captivated, likely by the bright artwork, fun graphics and animations. Although the puzzles get progressively more complex, my son made short work of the available levels, completing all of them within two sittings. It is possible to purchase additional levels, however, from the reviews, it would appear to be more of the same, not adding much to the learning experience.

It is worth noting that the levels are procedurally generated, therefore will automatically change if replayed. This is a great feature, but not necessarily varied or challenging enough to support long-term development.

With that said, as a free app, it is hard to not recommend!

Kodable

In recent weeks, my son has been working his way through another app called Kodable, which includes a comprehensive curriculum of software development challenges created specifically for kids.

The curriculum begins very much like Swift Playgrounds and Hopster Coding Safari, with drag and drop programming, leveraging symbols (arrows, etc.) Over time, it introduces additional concepts such as conditional code and loops, before transitioning to written code (either JavaScript or Swift).

The app consists of six main sections, which can initially make the main menu a little confusing, but once orientated, highlights the depth/breadth of content available.

Smeeborg: Solve puzzles using symbol-based code, designed for pre-readers.
Asteroidia: Begin your introduction to variables and data types.
Bug World: Transition from symbol-based code to written code in JavaScript or Swift.
Maze Maker: Practice foundational skills from every programming language as you create beautiful mazes.
Build Your Own Fuzz: Use creativity and imagination to practice your coding skills building your very own characters (known as fuzzes).
Game Designer: Design games by modifying properties of their code.

Each section includes a wealth of content, primarily delivered as chapters and levels which progressively increase in difficultly by introducing new programming concepts.

Starting with Smeeborg and recognising the similarities to Swift Playgrounds and Hopster Coding Safari, my son was quick to pick-up the concept and progressed through the first few chapters very quickly.

He did require occasional assistance, but the app does a nice job of providing visual cues if/when it detects the user has become stuck.

What makes this app stand-out to me is the gradual progression it offers, whilst always building upon skills previously demonstrated. For example, chapter three introduced “loops”, using visual cues that had been previously mastered in the previous chapters.

Each level offers a small reward, unlocking new parts for your avatar (known as a Fuzz), which the user can modify in the “Build your Own Fuzz” section. These small incentives, alongside occasional “bonus games” where you get the chance to win additional prizes, provides a fun way to encourage progression. My daughter (only two-years-old) enjoyed helping my son complete the “bonus games” and modifying the avatar.

Overall, I have been incredibly impressed with Kodable! However, my son is still working through Smeeborg, therefore I am assuming that the more advanced sections are delivered at the same quality.

Finally, Kodable is not free (which I think is fair, based on the content offered). It requires an annual subscription, which can be purchased for £59.99 but can often be found on offer for £49.99. Although not cheap, I feel there is enough content, delivered with enough polish that it is worth the investment.

Conclusion

Over the past five months, I have seen my son develop the core concepts associated with software development, whilst also improve his logical reasoning, maths, reading and writing.

He has also started to recognise the role of software within our world, often making statements about how something was “coded” or if he spots a potential bug.

Considering he also appears to enjoy the process (anything to play on the iPad), I plan to continue our journey into software development, likely sticking with Kodable, at least until he reaches a logical progression point (or gets bored).

I hope you have found this article useful. Below are three “getting started” recommendations.

Focus on core concepts, not a specific programming language or syntax.
Start with a free learning exerience (e.g. Apple Swift Playgrounds and/or Hopster Coding Safari, etc.)
If possible, look to introduce an option where the software interacts with a physical object (e.g. Sphero Mini, Lego, etc.)

Happy coding!

My Setup (Q3 2020)

Tue, 25 Aug 2020 00:00:00 -0500

Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

Home Setup

At home, I switch between four devices (three laptops and one desktop).

My daily driver is a Razer Blade 15 Advanced (Late-2019). It is primarily used for productivity tasks, software development, photo editing, virtual labs, gaming and game development (GameMaker Studio 2). The full specification can be found below:

Razer Blade 15 Advanced (Late-2019)
2.6GHz 6-Core Intel i7-9750H (Turbo up to 4.5GHz)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
1TB Samsung 970 EVO Plus PCI-e NVMe SSD
Intel UHD Graphics 630
NVIDIA GeForce RTX 2080 Max-Q 8GB GDDR6
15.6-inch 4K OLED Touch DCI-P3 HDR400 Display (3840x2160)

I also have a 16-inch Apple MacBook Pro (NOV-2019), which is a secondary device providing native access to the Apple ecosystem for software development, video editing, etc. The full specification can be found below:

16-inch MacBook Pro with Retina Display (NOV-2019)
2.4GHz 8-Core Intel i9-9980HK (Turbo up to 5.0GHz)
32GB 2666MHz DDR4 RAM
1TB PCI-e SSD
Intel UHD Graphics 630
AMD Radeon Pro 5500M 8GB GDDR6
16-inch IPS P3 Display (3072x1920, 500nits)

I have a 13-inch Dell XPS 13 (9300), which is reserved for Linux (specifically Fedora). Dell provides native support for Linux (primarily Ubuntu), making it a good candidate for Linux applications and testing. The full specification can be found below:

13-inch Dell XPS 13 (9300)
1.3GHz 4-Core Intel i7-1065G7 (Turbo up to 3.9GHz)
16GB 3733MHz LPDDR4x RAM
512GB M.2 PCI-e NVMe SSD
Intel Iris Plus Graphics
13.4-inch Display (1920x1200, 500nits)

Alongside my laptops, I have a custom-built desktop PC. The specification (listed below) is a couple of years old but still provides excellent overall performance. Although released in 2018, the GeForce RTX 2080 Ti is still a phenomenal graphics card, enabling high-performance Ray Tracing gaming.

Asus Maximus VIII Hero Alpha (Intel Z170)
Intel Core i7-6700K 4.0GHz @ 4.6GHz
Noctua NH-D15 CPU Cooler
32GB Corsair Vengeance LPX DDR4 PC4-24000C15 (15-17-17-35)
1TB Samsung PM981 M.2 PCI-e NVMe SSD
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

The Intel Core i7-6700K is four-years-old, however, Intel has provided little incentive to upgrade. For example, their upcoming 10th generation processors are still based on the Skylake microarchitecture, first released in 2015.

I continue to watch AMD closely, who are now the clear performance leaders with their Zen 2 microarchitecture, but will probably delay any future upgrades until transformative technologies such as PCI-e 4.0 are mainstream.

In 2020, I replaced my previous three monitors with one Super Ultrawide, specifically the Samsung C49RG90. The full specification of the monitor can be found below.

Model: Samsung C49RG90
Panel Size: 49-inch
Panel Type: VA
Panel Curvature: 1800R
Aspect Ratio: 32:9
Resolution: 5120x1440
Refresh Rate: 120Hz
Variable Refresh Rate: AMD FreeSync 2 (48-120Hz Range)
Response Time: 4ms (GTG)
Colour Accuracy: 125% sRGB, 92% Adobe RGB, 95% DCI-P3
High Dynamic Range: HDR1000
Brightness: 600cd/m2 (Typical), 1000cd/m2 (Peak)

My workspace is kept fairly minimal, the Razer Blade 15 Advanced, MacBook Pro and Dell XPS 13 can connect to the monitors via a StarTech Dual-DisplayPort to ThunderBolt 3 Adapter, providing a single-cable solution.

As you can see, for peripherals, I primarily use the Logitech Craft, Logitech MX Anywhere 3 and Logitech Brio Ultra HD Pro. The keyboard and mouse were selected as they include “Easy-Switch”, which makes it simple to toggle between multiple Bluetooth devices.

The image below provides a closer look at the Samsung C49RG90 monitor.

Additional peripherals include headphones (Bang & Olufsen BeoPlay H8), controller (Microsoft Xbox Elite Series 2), flight controller (Saitek X-55), steering wheel (Logitech G29) and virtual relaity headset (Oculus Quest).

As highlighted in the image below, the desktop PC is hidden under the desk, with the case door exposed providing easy access to the components.

The desk and matching pedestals are from Three Counties Office Furniture. They are designed for corporate use, therefore are hard-wearing and include integrated cable management.

Finally, I use a Herman Miller Setu chair, which includes the Kinematic Spine to control resistance and automatically support your weight as you recline.

Work Setup

At work, I use the Razer Blade 15 Advanced, Apple MacBook Pro and/or Dell XPS 13 (depending on the workload).

I use two 27-inch Dell U2718Q IPS HDR 4K monitors, connected via the previously mentioned StarTech Dual-DisplayPort to ThunderBolt 3 Adapter. To provided consistency, all other peripherals (e.g. Mouse, Keyboard, and Webcam) are the same as home.

Super Ultrawide

Mon, 24 Aug 2020 00:00:00 -0500

Over the years I have experimented with different monitor setups. I have tested almost every viable combination positioned for consumers, including:

Apple Cinema Display (High Resolution, Single Cable)
3D Compatible Monitor (3D Content)
Triple 24-inch Monitors (AMD Eyefinity for Gaming)
High Resolution (5K+) Monitor (Productivity)
High Refresh Rate (144Hz+) Monitor (Gaming)
Dual 27-inch 4K HDR Monitors (Massive 8K Resolution)

Every monitor setup had its relative strengths and weaknesses. For example:

High refresh rate monitors are great for gaming, but often trade-off colour accuracy and high dynamic range.
High-resolution monitors (4K+) provide great image quality, but are commonly locked at a 60Hz refresh rate, have a lower response time and require software DPI scaling.
Dual or triple monitor setups deliver an immersive experience as well as an amazing amount of screen real-estate but include cumbersome monitor bezels, as well as the need to connect multiple cables.

Over the past year, a new “type” of monitor has hit the market, known as the “Super Ultrawide”. These monitors include a massive 49-inch panel with a 32:9 aspect ratio, which is the equivalent of dual 27-inch monitors.

Interestingly, Super Ultrawide monitors have the potential to deliver the perfect balance between productivity and gaming, achieved through the following characteristics:

High Resolution: Up to 5120x1440, which is approximately one million pixels less than 4K (arguably the best resolution for Windows DPI scaling and game performance).
High Refresh Rate: 120Hz+ with Variable Refresh Rate (FreeSync, G-Sync, etc.)
Low Response Time: Maximum 4ms (GTG)
Colour Accuracy: 95%+ of DCI-P3 and HDR1000 (Peak Brightness 1000cd/m2).
Screen Real Estate: The equivalent of dual 27-inch QHD monitors.

As a result, I decided to replace my existing three monitors (details below), with a single Super Ultrawide.

2x Dell U2718Q IPS HDR 4K (3840x2160 / 60Hz / 5ms / HDR10)
1x Dell S2716DG TN G-Sync (2560x1440 / 144Hz / VRR / 1ms)

I have selected the Samsung C49RG90, which can be purchased for around £1000 and includes the following specification:

Model: Samsung C49RG90
Panel Size: 49-inch
Panel Type: VA
Panel Curvature: 1800R
Aspect Ratio: 32:9
Resolution: 5120x1440
Refresh Rate: 120Hz
Variable Refresh Rate: AMD FreeSync 2 (48-120Hz Range)
Response Time: 4ms (GTG)
Colour Accuracy: 125% sRGB, 92% Adobe RGB, 95% DCI-P3
High Dynamic Range: HDR1000
Brightness: 600cd/m2 (Typical), 1000cd/m2 (Peak)

The Samsung C49RG90 delivers in one monitor, the capabilities that I was previously achieving with three. Specifically, a 5120x1440 resolution, which is the equivalent of dual 27-inch QHD monitors, alongside a high refresh rate (120Hz) and support for AMD FreeSync 2 (which also is compatible with Nvidia graphics).

I did explore the new Samsung Odyssey G9, which includes a higher refresh rate (240Hz), alongside a 1000R curvature. However, there have been widespread reports on Reddit of inconsistent build quality (light leaking) and a “display flickering” issue. Therefore, considering I am not a competitive gamer, I decided to save some money (approximately £300) and select the “tried and tested” Samsung C49RG90.

Setup

Unsurprisingly the Samsung C49RG90 is a monster, something that is immediately apparent based on the size of the box. I had initially planned to leverage my existing monitor arm, however, the 11.6kg weight was simply to heavy for the single arm.

Therefore, I switched to the integrated stand, which consumes more desk space, but is certainly more robust. On a positive note, the stand does include an integrated cable tidy and handy headphone stand.

Setting up the monitor was very simple, my default settings can be found below:

Resolution: 5120x1440
Refresh Rate: 120Hz
Colour: 8-bit
FreeSync: Ultimate Engine (48-120Hz Range)
Brightness: 50 to 100
Contrast: 75
Sharpness: 56
Low Input Lag: On
Response Time: Fastest
Local Dimming: Auto

When viewing HDR content, I switch the Colour to 10-bit (Nvidia Control Panel) and Brightness to 100, which automatically lowers the Refresh Rate to 100Hz. I do not personally leave HDR on by default, recognising the current HDR limitations in Windows 10.

The Samsung C49RG90 supports FreeSync 2, which adds three important features to the FreeSync specification, specifically support for High Dynamic Range (HDR), Low Input Latency and Low Framerate Compensation (LFC).

In my opinion, Low Framerate Compensation (LFC) is the most important feature, as it essentially removes the minimum refresh rate threshold by automatically multiplying frames to ensure they exceed the minimum (e.g. 48Hz).

Although FreeSync is AMD technology, it works fine with Nvidia graphics. However, the setting “Enable settings for the selected display model” must be enabled in the “Set up G-Sync” tab of the Nvidia Control Panel.

Usage

The 32:9 aspect ratio is simply phenomenal, delivering a truly immersive experience regardless of the task. The 1800R curvature works great for productivity (minimal distortion), but I could see the benefits of a 1000R curvature, especially when playing games.

Unlike having dual displays, application windows running on the Super Ultrawide can be positioned in the centre of the display, something that would have been previously blocked by the monitor bezels.

This subtle change makes a big difference, especially for focused tasks such as coding, writing and/or reviewing documents. Where previously I would have my primary application window on the left display, I now position centrally, with any “secondary” application windows positioned either side.

Another key advantage of a single monitor is that applications can leverage the full screen by default. Although not appropriate for all use cases (a single browser window across 32:9 aspect ratio looks ridiculous), applications such as Zoom deliver an experience that could only be achieved on this setup.

My only negative feedback would be related to text clarity, which is a common complaint with monitors that use a VA panel. The issue is most noticeable when compared against a high-resolution IPS panel, but is not something that bothers me in general use. It is worth noting that text clarity can be tuned in Windows 10 using the ClearType settings.

This immersive video conferencing experience has been particularly useful in 2020, where due to COVID-19, I have spent a lot of time and home.

Gaming is another highlight of the Samsung C49RG90, with the 120Hz refresh rate and FreeSync 2 delivering a responsive, consistent, visual experience.

With that said, not every game natively supports the 32:9 aspect ratio, therefore it is worth checking before any purchase. I would recommend starting with the “Steam Ultrawide Master Race” group, which includes a curated list of supported games.

Assuming support, the Super Ultrawide experience is breathtaking! I have included a few photos below, which hopefully highlight the immersion.

Red Dead Redemption 2 is a first and third-person perspective open-world game developed by Rockstar Games. The open landscapes look stunning when running at a 32:9 aspect ratio. When playing first or third-person games on a Super Ultrawide, the Field of View (FoV) might require adjustment, something that can usually be done from within the game options.

Another recent release is Flight Simulator from Microsoft, which includes native support for the 32:9 aspect ratio. Flight Simulator is a very demanding game, therefore does require impressive hardware to run at the native 5120x1440 resolution (although this resolution is still less demanding than 4K).

Finally, I am a fan of racing games (specifically the F1 Series from Codemasters), which I have always enjoyed playing with ultrawide setups (specifically AMD Eyefinity, as highlighted in the photo below).

As highlighted in the photo below, the impact of Super Ultrawide monitor is immediately evident, simplifying the setup and removing the bezels.

Although not perfect for every gaming experience, I enjoy the Super Ultrawide gaming experience. However, this assumes a PC that can push a 5120x1440 resolution at a respectable framerate.

Conclusion

In short, I believe am pleased with my switch to Super Ultrawide. The Samsung C49RG90 delivers upon the promise of high resolution, high refresh rate, with minimal compromise.

Dell XPS 17

Sun, 02 Aug 2020 00:00:00 -0500

In June, Dell released the new Dell XPS 17 laptop, which included a refreshed design aesthetic and an impressive 17-inch display.

Over the years, I have experimented with different display sizes, looking to find the right balance between performance and portability. This continuous search has always led me back to 15-inch laptops, which tend to be the entry point for high-performance processors and graphics.

Traditionally, I have used a 15-inch MacBook Pro, but I am currently using a Razer Blade 15 Advanced as my daily driver.

Considering the form factor, the Razer Blade 15 Advanced offers excellent performance, packing a full NVIDIA GeForce RTX 2080 with Max-Q, whilst still competing with the Apple MacBook Pro for size and aesthetics.

However, the design of the 15.6-inch display included on the Razer Blade 15 Advanced is starting to look dated, likely due to the 16:9 aspect ratio, which results in a large bezel at the bottom (often described as the chin).

This is where the new Dell XPS 17 shines! Dell has successfully included a 17-inch display, with a 16:10 aspect ratio, delivered in a laptop that is only fractionally larger than most 15-inch counterparts.

Dell has achieved this feat by reducing the display bezels to only a few millimetres (known as InfinityEdge), achieving an impressive 93.7% screen-to-body ratio.

The result is so striking that it immediately makes other laptops look dated, including the recently refreshed Apple MacBook Pro, which attempted the same trick with a 16-inch display (84% screen-to-body ratio).

As a result, I could not resist testing the new Dell XPS 17.

This article is not a review (there are many online), simply a summary of my experience using the laptop, specifically compared against my Razer Blade 15 Advanced and Apple MacBook Pro.

Design

I would ordinarily start with the laptop specification, however, the real star of this show is the display and new design aesthetic.

The photo below highlights the sheer scale of the display (you can just about see the Windows “Start” button for reference).

The 16:10 aspect ratio means that the display seamlessly fades into the body, delivering more vertical real estate when compared with most laptops (such as my Razer Blade 15 Advanced).

I selected the UHD+ (3840 x 2400) touch display, which is very good but due to the use of an IPS panel, does not quite deliver the black-level performance and visual punch offered by the Razer Blade 15 Advanced, which uses an OLED panel. With that said, the display is very colour accurate and therefore a pleasure to use for productivity use cases (I will mention gaming later in the article).

Outside of the display, the new Dell design aesthetic (also found on the new XPS 13 and 15) is stunning! The body is dense and rigid, with clean lines and a build-quality that rivals Apple.

The carbon-fibre palm-rests, which have become a consistent feature across the XPS range, continue to act as a design differentiator, whilst also delivering a “softer” feel when compared against aluminium.

Additionally, as someone who wears a steel watch bracelet, I can confirm that the carbon-fibre palm rests are also more durable than aluminium, being more resistant to scratches and discolouration (my Razer Blade 15 Advanced and Apple MacBook Pro both have a thin silver line on the edge of the palm rest).

The photo below shows the Dell XPS 17 (bottom) stacked with an Apple MacBook Pro 16-inch, Lenovo ThinkPad X1 Extreme 15-inch and a Razer Blade 15 Advanced 15-inch (top).

As you can see, Dell has successfully packed a 17-inch display into a body that is only fractionally larger than competing 16-inch and 15-inch laptops. The difference in size is so small that it still comfortably fits into my medium Timbuk2 Messenger Bag, which was designed for a 15-inch laptop.

Although these are impressive dimensions, the Dell XPS 17 is obviously a large laptop. It is also heavy, weighing 2.51kg, compared to 1.9kg of the Apple MacBook Pro 16-inch. I did not find the additional size or weight to be an issue but accept that this will not be the case for everyone.

Finally, the keyboard and trackpad are both excellent, only rivalled by the Apple MacBook Pro and Razer Blade 15 Advanced (Apple are still the king of the trackpad). I did notice some minor “wobble” on the trackpad (also reported in some reviews), but nothing that I would consider a deal-breaker.

In short, I consider the Dell XPS 17 design close to perfect!

Performance

The Dell XPS 17 is available in a range of specifications, starting at £1799. The model I selected was £2899, which included the UHD+ display and NVIDIA GeForce RTX graphics.

Dell XPS 17 (9700)
2.3Hz 8-Core Intel Core i7-10875H (Turbo up to 5.1GHz)
16GB DDR4 2933MHz RAM
1TB M.2 NVMe SSD
Intel UHD Graphics 630
NVIDIA GeForce RTX 2060 with Max-Q 6GB GDDR6
17.0” UHD+ InfinityEdge Touch Display (3840 x 2400, 500nits)

Like the majority of laptops in 2020, the Dell XPS 17 comes equipped with an Intel CPU. Although perfectly adequate, the recent release of the AMD Ryzen 4000 Series Mobile Processors, would have made for a fascinating inclusion. However, similar to the Apple MacBook Pro, the Dell XPS 17 has a heavy reliance on Thunderbolt 3 for I/O (4x ports), which for all intents and purposes, is still Intel technology.

Ideally, I would have configured the laptop with 32GB RAM, however, for some reason, this option was not available in the UK (only the US). Thankfully, the Dell XPS 17 can be easily upgraded, including the RAM (up to 64GB) and SSD (2x M.2 nVME) (assuming you can remove the Torx screws, without stripping them).

The NVIDIA GeForce RTX 2060 with Max-Q offers good performance, making the Dell XPS 17 viable for gaming. However, this is not a gaming laptop, which is emphasised by the fact that Dell pre-install the NVIDIA Studio Driver, which is designed for artists, creators and 3D developers. I proceeded to install the NVIDIA Game Ready Driver (GRD), which generally provides better performance for my workloads.

Overall, performance is solid (in line with what you would expect from this specification). The table below highlights a few benchmark results, comparing the Dell XPS 17 against my Razer Blade 15 Advanced and Lenovo ThinkPad X1 Extreme.

I did not complete detailed testing of the thermals but did not notice any major concerns. The Dell XPS 17 is equipped with a vapour-chamber, covering the CPU and Graphics, which should be adequate for this specification.

Finally, one minor disappointment. Dell has chosen to charge the laptop via the USB-C port, which although is incredibly convenient, does not deliver enough power (maximum 130W) to charge the laptop whilst running at full load. As a result, the laptop will continue to draw power from the battery, even when plugged in (not ideal).

I would have prefered a dedicated charging port with adequate power delivery (like the Razer Blade 15 Advanced), whilst also supporting USB-C charging for added convenience.

Conclusion

In summary, the Dell XPS 17 is a phenomenal laptop!

Dell has nailed the 2020 design aesthetic, with a unique, high-quality unibody aluminium body, paired perfectly with the carbon-fibre palm-rests.

The display is the most impressive part of the laptop, with the 16:10 aspect ratio and 93.7% screen-to-body ratio. Working from the Dell XPS 17 has been a pleasure, especially productivity activities (multiple windows) and software development. The size and aspect ratio of the display delivers almost “desktop” levels of screen real estate, all within a package that is still viable for travel. This is a remarkable accomplishment!

With that said, would I replace my Razer Blade 15 Advanced with the Dell XPS 17? At this point, no. However, only because I favour the OLED display and higher-performance graphics. If the Dell XPS offered these options, alongside a dedicated charging port, I would be very tempted!

In conclusion, I consider the Dell XPS 17 an easy product to recommend, especially for those looking to switch from an Apple MacBook Pro, as I consider the Dell XPS range the closest equivalent to what Apple offers (balancing style and performance).

Next.js

Fri, 24 Jul 2020 00:00:00 -0500

Next.js is a React framework, built on Node.js, that is rapidly gaining popularity amongst web developers.

Next.js is an open-source project, which is maintained (including the copyright and trademarks) by Vercel.

The goal of Next.js is to provide the building blocks to create fast web applications, complementing the React (Facebook-backed) JavaScript library.

React is an excellent library for building interactive web applications, providing helpful functions to build user interface components. However, it does not stipulate or facilitate where these functions are used within the application.

Next.js handles the tooling and configuration not covered by React, providing additional structure, features, and optimisations.

As a result, Next.js is benefiting from the popularity of React, allowing developers to start with React, and then incrementally adopt Next.js to help solve common challenges such as routing, data fetching, integrations, etc.

Next.js also supports Incremental Static Regeneration (ISR), which uses static generation on a per-page basis, removing the need to rebuild the entire site, as well as Static Site Generation (SSG), which is an approach I encourage, following a JAMstack architecture.

My personal experience with React is fairly limited, however, I do work with React development teams and am already seeing the huge potential with Next.js. Therefore, if you are a React developer, I recommend you review the Next.js getting started documentation.

HoloLens 2

Mon, 20 Jul 2020 00:00:00 -0500

Over the past few weeks, I have been testing the Mirosoft HoloLens 2.

Released in November 2019, the HoloLens 2 is a mixed reality (MR) headset, which aims to produce environments and visualisations, where physical and digital objects co-exist and interact in real-time.

The mixed reality (MR) terminology can be confusing, especially when discussed alongside virtual reality (VR) and augmented reality (AR). The descriptions below outline the key differences.

Virtual reality (VR) fully immerses the user in an artificial digital environment.
Augmented reality (AR) overlays virtual objects on the real world.
Mixed reality (MR) goes beyond augmented reality, anchoring virtual objects on the real world.

As highlighted in the articles “Virtual Reality” and “Oculus Quest”, computer-generated simulations have matured significantly over the past five years. We have not yet achieved the fidelity of the Star Trek Holodeck or Star War Holograms, but these science fiction concepts no longer feel unachievable.

When the HoloLens was first introduced in 2016, Microsoft failed to position the product effectively, setting a false expectation that it would be made available as a consumer product, similar to Virtual Reality Headsets (e.g. Oculus and Kinect.

Thankfully, Microsoft has rectified this confusion with HoloLens 2, clearly positioning the product for enterprise businesses, supporting use cases such as remote support, 3D modelling, training, hands-free procedures, etc.

As an example, the video below highlights HoloLens 2 being used alongside Dynamics 365 Remote Assist for hardware remote support.

Over recent months, the applicability of these use cases has been amplified, recognising the global restrictions being applied due to the COVID-19 pandemic.

In this article, I plan to share a brief insight into my testing (results to following in a future article), as well as my initial impressions of the HoloLens 2 hardware and software.

Testing

I have been partnering with World Wide Technology (WWT) to test two use cases, which have been documented in the article “Virtual Migration Strategy with Microsoft HoloLens 2”. A summary of the use cases can be found below.

Deployments and Site Migrations

Leverage the HoloLens 2 to assist physical hardware deployments and site migrations, specifically applicable to R&D and Manufacturing sites.
Instead of having a large number of expert resources on-site, a smaller number of local resources (known as “smart hands”) leverage HoloLens 2 to execute the required processes.
Local resources use the HoloLens 2 and Dynamics 365 Guides, which delivers a “digital playbook”, providing a fully annotated step-by-step guide.
Expert resources provide remote consultancy and support via Dynamics 365 Remote Assist, including real-time voice, video and annotation.

Regulated Testing and Validation

As part of a regulated industry, the qualification process includes visual validation of the IQ (installation qualification) and PQ (performance qualification) testing.
Evidence must be produced and reviewed by a quality assurance resource, confirming the system configuration/installation has been completed in a controlled fashion.
Local resources use the HoloLens 2 and Dynamics 365 Remote Assist to connect with remote quality assurance resources.
The HoloLens 2 cameras could be used to record all testing, including human interactions with annotations. The recording can be retained as documented evidence of the testing process.

The hope is that the HoloLens 2 and associated services can help improve the quality and efficiency of these use cases, reducing the need for on-site resources (travel and costs), whilst improving speed to value and providing a greater level of agility via real-time collaboration.

Hardware

At first glance, the compute, storage and network capabilities of the HoloLens 2 are fairly underwhelming, essentially equivalent to a mid-range smartphone.

SoC: Qualcomm Snapdragon 850
Memory: 4GB LPDDR4x System DRAM
Storage: 64GB UFS 2.1
Connectivity: Wi-Fi 5 (802.11ac 2x2), Bluetooth 5, USB Type-C

The Qualcomm Snapdragon 850 Mobile Compute Platform is an ARM-based SoC designed for Windows laptops. It is essentially a higher clocked version (3.0Ghz vs. 2.8GHz) of the popular Snapdragon 845 Mobile Platform, manufactured using 10nm LPP FinFET at TSMC.

Thankfully, the HoloLens 2 is more than just a smartphone that you mount on your head. It also includes a unique display, a wide range of sensors, as well as a custom multiprocessor known as the Holographic Processing Unit (HPU).

It is the HPU that makes the HoloLens 2 unique, removing the need for a cumbersome breakout box and cables, which would severely limit real-world adoption.

The HPU is responsible for processing the information coming from the sensors, including the custom time-of-flight depth sensor, head-tracking cameras, inertial measurement unit (IMU) and infrared camera.

HoloLens 2

As previously stated, the HoloLens 2 is not targetted at the consumer market. However, the packaging is certainly premium quality, that would not look out of place being sold at an Apple Store. This care and attention regarding the “unboxing experience” is unusual for an enterprise product, likely highlighting that the HoloLens 2 is still highly specialist (limited audience).

Inside the box, the HoloLens 2 includes a high-quality travel case (very similar to the optional Oculus Quest Travel Case), which stores and protects the headset itself, as well as the USB-C power adapter.

The HoloLens 2 headset is very well built, with premium materials and tight tolerances. The product feels rigid and dense, which promotes the quality, but also practical considering the target market.

In the Virtual Reality world, the “halo” head strap (found on the PSVR and Oculus Rift S) is considered the most comfortable. However, Microsoft has chosen to implement a traditional velcro headband, with the addition of an “adjustment wheel” at the rear, which can be used to quickly and easily tighten/loosen the headband itself.

This “hybrid” design is very comfortable and should be suitable for almost any head size (including my five-year-old son, as shown in a later photo).

At 566g, the HoloLens 2 weighs about the same as the Oculus Quest (571g), but the weight is more evenly distributed, with the battery being located at the back of the headset. This delivers a very balanced product, making it more viable for prolonged usage.

Wearing the HoloLens 2 is surprisingly comfortable, although not subtle (you cannot hide the fact you are wearing the headset). Therefore, it is likely a “point in time” device, only worn when completing a specific task.

When viewed from the front, The HoloLens 2 looks a little like wearing a large bicycle helmet and cycling sunglasses.

Thanks to the open top and sides, the HoloLens 2 does not feel constrained or claustrophobic, which can be a common criticism of Virtual Reality headsets.

Although the lenses are transparent, the embedded display can obscure vision. As a result, Microsoft has implemented a flip mechanism, allowing the lenses to be moved from the field of view. This process awkwardly shifts the weight of the headset but is still useful for quick conversations outside of mixed reality.

Overall, I am incredibly impressed with the HoloLens 2 hardware. Although the technology is relatively new and likely targeting a very specific audience, these limitations are not reflected in the design and/or build quality.

In short, Microsoft has clearly invested the time to deliver a high-quality “production-ready” product, instead of hiding behind common nomenclature such as “Developer Edition”, which is often used as an excuse for reduced quality and/or missing features.

Software

The HoloLens 2 runs Windows Holographic OS, which is part of the Windows 10 family.

Like all other versions of Windows 10, it includes a Start menu, which can be triggered via a one-handed gesture (pinching motion) or by gazing at your wrist, which triggers the Windows logo to appear.

Alongside the Start menu, other common Windows concepts are sprinkled throughout the operating system, specifically the Application Launcher, Edge Browser, Mail, Calendar, Store, etc. The HoloLens 2 is also fully AutoPilot and InTune compatible, meaning the device can be managed centrally like any other Windows 10 endpoint. These management features, alongside a valid Microsoft Account, helps to ensure the setup process is quick and simple, providing instant access to user applications, data, etc.

Recognising the lack of a standard input device (e.g. Mouse and keyboard), user interaction is achieved via “Gestures”, “Gaze”, “Hand Ray” and voice.

Gestures allow the user to interact with virtual objects using their hands and fingers.
Gaze allows the user to interact with virtual objects by looking (gazing) at specific items (e.g. Application, Buttons, etc.)
The Hand Ray presents the user with a “laser pointer”, enabling gestures to be used on virtual objects that are anchored out of direct reach.
The HoloLens 2 includes a microphone and speaker array, therefore can accept voice commands as an input (similar to voice assistants such as Amazon Alexa).

Gestures, gaze and voice can be used simultaneously, delivering a reasonably intuitive experience, with the hand ray reducing the need for physical movement (a common challenge in mixed reality, where virtual objects are anchored in the real world).

Similar to when multi-touch was first introduced, the new gestures do have a learning curve. For example, to “air tap”, the user must:

Create a loose fist, pointing the index finger straight up.
Tap the finger down.
Quickly raise it back up again.

Navigating the operating system and applications is fairly simple, primarily accomplished via air taps. For example, the image below is my five-year-old son interacting with Holo Lens 2. I had previously hidden multiple virtual dinosaurs around our house, which he then had to locate using mixed reality. Thanks to the advanced anchoring of virtual objects, the virtual dinosaurs look and feel part of the physical world. His reaction to seeing the dinosaurs hiding around our house was priceless and demonstrated the “magic” of this technology.

At the time of writing, there are 351 HoloLens 2 compatible applications in the Microsoft Store, highlights include 3D Viewer, Galaxy Explorer, Graffiti 3D, HoloBlocks, HoloTour, Robo Raid, HoloAnatomy, etc. Unsurprisingly, the majority of the applications are still very experimental, offering a glimpse into the “art of the possible”, but rarely adding tangible value.

Unfortunately, activities that require traditional typed input are far more cumbersome. Although Windows Holographic OS includes a virtual keyboard, the lack of any tactile feedback makes the process of typing very slow and prone to error. To make matters worse, the requirement to hold your arms up without a palm rest is surprisingly difficult, which amplifies the issues. I assume, as the application ecosystem matures, other interaction mechanisms, such as voice, will be favoured.

Unsurprisingly, HoloLens 2 is at its best when it is operating in true mixed reality mode (virtual objects anchored on the real world), not when attempting to mimic traditional productivity activities. If you need to write a document or respond to e-mail, your laptop is still a far better choice.

The video below was taken directly from HoloLens 2 and highlights me interacting with the operating system and a few applications. I have purposely not edited, enhanced or refined the video, hoping to highlight the good, the bad and the ugly. You will see me (more than once) fail to interact with a virtual object on the first attempt.

As you can see, the virtual overlay and anchoring of virtual objects is incredibly impressive and certainly beyond any other mixed reality or augmented reality device I have ever used. However, you can also see how the interactions can be cumbersome, which can break the illusion, resulting in frustration.

Conclusion

In conclusion, I am very impressed by the HoloLens 2 and excited to see how the product evolves. I am pleased Microsoft remain willing to invest in these innovative technologies, especially considering past failures and accepting that the viable market is very small (at least today).

Microsoft should also be commended for the quality of the product, which feels “production-ready” from the moment you pick-up the box. The hardware feels refined and robust (ready for business) and while the software needs some refinement, it is surprisingly intuitive and usable.

Finally, I am pleased that Microsoft targetted and achieved basic impact protection requirements (ANSI Z87.1, CSA Z94.3 and EN 166), making the product viable for my test use cases. Over the coming weeks, I will conclude my testing, therefore I look forward to seeing how HoloLens 2 performs, as well as the reception it receives from the users. I’ll be sure to post an update with the results.

Coding Stats

Wed, 01 Jul 2020 00:00:00 -0500

I recently reviewed my statistics on CoderStats, which pulls public information from the GitHub API. My statistics are fairly underwhelming, but (I assume like many others) they do not tell the full story.

I have always worked in IT, initially as an analyst, prominently in infrastructure roles (web hosting, networking, etc.) I eventually progressed into architecture, working alongside engineering teams. Although I have never had a role as a dedicated software engineer, the rise of software-defined techniques and technologies, alongside the growth of digital business models, presented opportunities to get “hands-on keyboard” and code.

Alongside work, I have always had a strong personal interest in software development (mostly web technologies) and have explored a lot of different programming languages, specifically JavaScript, Ruby, PHP, Python, Apex, GML, etc.

Over the years, I have written numerous articles highlighting my coding projects.

2011: PhoneGap
2013: Wordpress on Heroku
2014: IE Rendering Modes
2014: VF / IE Rendering Issue
2014: Force.com Application Platform
2014: JavaScript - Part One
2014: JavaScript - Part Two
2014: Developing on Force.com
2014: JavaScript - Part Three
2015: Installing Rails on OS X
2015: Jekyll on Heroku
2015: ES6 - Part One
2015: Conf Buddy
2015: Force.com Standards
2015: ES6 - Part Two
2015: Less with Grunt
2016: Auth0 and Rails
2016: Materialize and Force.com
2016: Node Architecture
2016: Site Builder
2016: Harp and Heroku
2016: GraphQL
2016: Site Builder
2016: Python Cheat Sheet
2016: Processing (p5.js)
2017: GameMaker Studio 2
2017: Alfred Snippets
2017: App Cloud - Hybrid Apps
2017: Jekyll and Docker
2017: Blockchain Playground
2017: Progressive Web Apps
2017: Blockchian Playground UI
2017: JAMstack
2018: Jekyll PWA
2019: Tank Arena
2019: Swift Playgrounds
2020: iPad Development

Due to the nature of my work and personal projects, the majority of my coding is private (stored in private repositories). Therefore, sites like CoderStats are unable to show a holistic view.

As a result, I thought I would share my full statistics for GitHub, including public and private contributions.

2016: 671 Contributions
2017: 81 Contributions
2018: 321 Contributions
2019: 221 Contributions
2020: 372 Contributions

NOTE: Before 2016, my work used Atlassian Bitbucket as their SCM system, which I also used for personal projects (providing consistency). Unfortunately, when I switched to GitHub in late 2015, I did not retain my previous contributions.

In reality, my full statistics are still not particularly impressive, but certainly demonstrate higher engagement than my public persona.

The year 2016 was stand-out for me, where I was the lead developer on two enterprise web applications, delivered as part of an extra-curricular work assignment. I showcased these web applications in the articles “Conf Buddy” and “Site Builder”.

In 2017, I contributed to several Blockchain projects, focused on education (e.g. Hash Trees, Consensus Algorithms, etc.) Specifically, Blockchain Playground and Blockchian Playground UI.

Regarding development language rankings, my private contributions align with my public persona, with a strong preference for JavaScript. CoderStats flags “Yacc” (Yet Another Compiler-Compiler), which is actually “GML” (GameMaker Language).

JavaScript
HTML
PHP
Ruby
Visualforce
Apex
Yacc (GML)
CSS

Unsurprisingly, as my responsibilities at work (and home) have increased, my coding time has naturally decreased. However, even in my role as Chief Technology Officer (CTO), I believe maintaining tangible skills is of critical importance. Therefore, I aim to explore one or two projects per year, generally targetting new technologies and/or new techniques that force me to stay “current” with the latest trends. For example, Apple Swift and Progressive Web Apps, etc.

Finally, I am increasingly interested in supporting initiatives that promote coding with children and young adults, either through coding clubs, hackathons and coding applications/services. This includes exploring and creating coding challenges that I complete with my children.

WSL 2

Tue, 12 May 2020 00:00:00 -0500

In my previous articles “WSL” and “VMware Workstation Pro”, I described my experience running Linux on Windows to enable a POSIX compatible development environment.

I leverage a combination of Windows Subsystem for Linux (WSL), Linux Virtual Machines running on VMware Workstation Pro and Docker. Unfortunately, I have yet to find a perfect workflow, with each technology introducing different limitations regarding performance, battery life, interoperability, stability, etc.

With this in mind, I have been eagerly anticipating the general availability of Windows Subsystem for Linux 2 (WSL 2), which for the first time brings a native Linux kernel to Windows. On paper, this mitigates many of my concerns, whilst also removing the need for heavy-weight virtualisation technologies to run Docker, etc.

In the annoucement blog post, Microsoft made some bold claims regarding potential performance improvements.

“Initial tests that we’ve run have WSL 2 running up to 20x faster compared to WSL 1 when unpacking a zipped tarball, and around 2-5x faster when using git clone, npm install and cmake on various projects.”

I hope these performance improvements translate to my build process, specifically when using Docker.

This article will cover the WSL 2 installation process, as well as the results from my initial testing.

Installing WSL 2

To get started, ensure you are running Windows 10 2004 (AKA 20H1). If not, it can be downloaded via Windows Update and/or the “Download Windows 10” website.

Although not required, I would also recommend installing the new Windows Terminal, which offers native WSL support.

Once ready, open PowerShell as an administrator and run the following commands.

dism.exe /online /enable-feature /featurename:Microsoft-Windows-Subsystem-Linux /all /norestart

dism.exe /online /enable-feature /featurename:VirtualMachinePlatform /all /norestart

Restart your system, allowing the installation of WSL and the Windows Virtual Machine Platform to complete.

Install the latest version of the WSL 2 Linux Kernal from the Microsoft website using the link below.

https://wslstorestorage.blob.core.windows.net/wslblob/wsl_update_x64.msi

Next, re-open PowerShell as an administrator and set WSL 2 as the default using the following command.

wsl --set-default-version 2

WSL 2 is now setup, meaning it is time to install a Linux Distrbution from the Windows Store. There are many options (Ubuntu, Fedora, Debian, OpenSUSE, Kali), but I generally select the latest LTS release of Ubuntu. At the time of writting this is Ubuntu 20.04.

Once installed, you will be prompted to create a username and password for the Linux distribution.

Finally, it is worth confirming that your new Linux distribution is running WSL 2. This can be achieved by running the following PowerShell command as an administrator.

wsl --list --verbose

If the version of WSL is incorrect, it can be easily modified by running the following PowerShell command as an administrator.

wsl --set-version <distribution name> <versionNumber>

You can now access WSL 2 by opening the Windows Terminal and selecting your Linux distribution (e.g. Ubuntu 20.04) as a new tab.

Testing WSL 2

With WSL 2 up and running, I installed the latest version of Docker Desktop and confirmed that it was configured to utilise WSL 2 as the backend.

I launched Docker Compose, which is configured to automatically build my application services, including the runtime environment (Ruby) and Static Site Generator (Jekyll).

Running Docker on Hyper-V with Windows 1909, the Jekyll build process would take on average 175 seconds to complete. Outlined below was the result for WSL 2.

As you can see, not what I had hoped. To rule out a one-time anomaly, I re-ran the build process several times, achieving an average of 503 seconds.

At this point, I knew I had something configured incorrectly and hit the forums.

Thankfully, the answer was fairly obvious, I had cloned my Git repository on the Windows filesystem, which meant that WSL 2 was using a mount point to access the files. This approach has a significant I/O performance impact, actually worse than running Docker on Hyper-V.

To remediate, I cloned my Git repository again, this time natively within the Linux filesystem. The improvement was immediate and significant.

At 73 seconds, the build process is now over 50% faster than Docker on Hyper-V with Windows 1909.

Thanks to the new Windows Terminal, accessing and manipulating files within the Linux filesystem is very simple. You can also quickly jump to an explorer view if you prefer a GUI. Simply navigate to the required folder and run the following command.

explorer.exe .

This will open a Windows Explorer at the desired location.

Conclusion

Overall, I have been impressed with WSL 2. Not only does it demonstrate Microsoft’s commitment to Linux on Windows, but also makes the developer experience comparable to using macOS and/or a native Linux distribution.

This is an impressive outcome, arguably positioning Windows as the most versatile operating system for developers.

IT Ecosystem

Tue, 12 May 2020 00:00:00 -0500

In January 2019, I started a series of articles that explored the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

The articles predominantly focused on the architecture, technology and positioning of key capabilities, but also (where appropriate) referenced facilitating processes and methodologies.

Recognising that all good IT decisions must include business context, I defined the following hypothetical business characteristics.

Global Business (Americas, EMA, APAC)
Regulated Industry (Food and Drug)
Customers in 90+ Countries
Value Chain includes R&D, Manufacturing, Commercial and Global Services
15,000+ Total Users (Employees and Contractors)
170+ IT Users (Employees and Contractors)
50+ Physical Sites Globally
10+ R&D and Manufacturing Sites (24x7 Operation)
Established Traditional Business Model
Emerging Digital Business Model
$5Billion Revenue

I used these characteristics to inform my proposed architecture, ensuring the content was grounded by a “real-world” scenario, hopefully making it relatable and credible.

Up to now, I have written twelve articles (listed below), covering a wide range of topics. Although each article covers a specific area, they are best reviewed in order, starting with “Modern IT Ecosystem”, which provides the required.

Part One: Modern IT Ecosystem
Part Two: Service Delivery
Part Three: SD-WAN
Part Four: Hybrid Multi-Cloud
Part Five: Automation
Part Six: Device-as-a-Service
Part Seven: Identity Access Management
Part Eight: Unified Communications
Part Nine: Line of Business Apps
Part Ten: Architecture Community
Part Eleven: Data Protection
Part Twelve: SIAM
Part Thirteen: IT Ecosystem

This final article (part thirteen) aims to conclude the series (at least for now), providing a summary of the decisions, which contributed to my vision of a Modern IT Ecosystem.

Modern IT Ecosystem

In the article “Modern IT Ecosystem”, I shared the following “marchitecture” diagram, which I aimed to highlight my desired end-state.

Throughout the past year, I have essentially populated this diagram, detailing the architecture, positioning technologies, as well as highlighting the dependencies and facilitating processes.

The diagram below acts as a conclusion, overlaying the technologies and vendors that I have proposed throughout the series. The details and rationale for each decision can be found within the individual articles.

As you can see from the diagram, with my previously described business characteristics in mind, a modern IT ecosystem can quickly become a complex landscape, supported by a wide range of technologies and vendors. With that said, I believe the philosophy, principles and declarations outlined in the article “Modern IT Ecosystem” provided a strong foundation for decision making, helping to drive synergy, improve efficiency and reduce technical debt.

Although I feel confident about my proposed architecture, there are many areas and details I simply did not have the time to document. For example, areas such as Domain Name Service (DNS) provider, Public Key Infrastructure (PKI), Privileged Access Management (PAM), etc. These areas (alongside many others) support critical capabilities that must exist to enable a secure, scalable IT ecosystem.

It is also worth noting that a lot can change within a year, therefore technologies positioned as part of my proposed “modern” IT ecosystem have already been superseded, etc. For example, I had positioned RSA Identity Governance and Access (IGL) as part of my proposed Identity Access Management architecture. Although still a perfectly viable option, part of my decision was driven by the desire to maximise the investment with Dell Technologies. In February 2020, Dell Technologies announced their intent to divest RSA, which would reduce the value proposition associated with my decision.

Cloud is another area that continues to evolve quickly, with new capabilities being added at a furious pace. For example, Microsoft continues to expand and enhance services within Office 365 and Azure, which could potentially be used to replace third-party solutions, enabling a better user experience and reducing costs (especially if bundled under the Microsoft 365 E5 License).

Conclusion

Overall, I hope the “thought experiment” documented throughout this series has provided a useful perspective regarding the creation of a modern IT ecosystem.

I do not claim that my proposed architecture is authoritative (there is an infinite number of options), especially considering all IT decisions must be made in the context of a specific business outcome, with appropriate situational awareness.

With that said, recognising the previously defined business characteristics, I believe this series has accomplished my goal of establishing a robust and highly flexible IT ecosystem for a traditional enterprise business, with emerging digital business opportunities.

For example, prioritising modern technologies and techniques, such as the use of Public Cloud, following an API-Centric Architecture that is Software-Defined and highly automated. This positioning, combined with pragmatic architecture decisions (e.g. Colocation Data Centres) help ensure the proposed IT ecosystem could support almost any workload (legacy or modern).

Finally, it is important to recognise that although this series has predominantly focused on architecture, this is only one piece of the puzzle, with the processes, methodologies and organisation design being critical when looking to unlock the total value proposition. I covered a few of these areas, but would likely need to write a separate series to provide the required level of detail (maybe something for the future).

iPad Development

Sat, 02 May 2020 00:00:00 -0500

In my article “iPad for the Enterprise (2020)”, I shared my workaround to enable offline (local) software development on the iPad. This involved the use of a Raspberry Pi 4, acting as a portable Linux server running Docker.

This article will explain how I achieved this setup, which is not something I would necessarily recommend, but was a fun little technology project.

First, why is this workaround even required? Unfortunately, due to Apple iPadOS restrictions, it is almost impossible to set up, customise and run a local development environment (e.g. Node.js, Ruby) on the iPad. Some applications do let you execute code locally (e.g. Pythonista), but these are often limited to prototyping, automation or simple code snippets.

The simplest workaround to this restriction would be to connect to a remote development environment, for example, AWS, Google Cloud, Azure, Heroku, etc. However, this would require a persistent Internet connection, which is not something I can always guarantee (especially when travelling). Therefore, I desire a local development environment that can be used offline, without an external power source.

When developing on the iPad, I use the following applications:

Git Client: Wroking Copy
Code Editor: Coda
Test Editor: iA Writer
SSH: Termius

When combined, these applications enable a reasonable developer workflow (still limited by PC standards).

Clone git repository from GitHub using Working Copy.
Access the cloned repository via Apple Files.
Modify the required files (Coda and/or iA Writer).
Push the changes back to GitHub using Working Copy.
Trigger the remote build process (if required) using Termius.

My frustration was step 5, the remote build process, which is what I mitigated through the use of a Raspberry Pi 4, acting as portable Linux server running Docker.

The Raspberry Pi 4 is well-positioned for this job, as it is small, light and cheap, but also very capable (Quad-Core CPU, 4GB RAM, SD-Card Reader). Thanks to the USB-C port, it is also possible to set up a physical network connection and power the device directly from the iPad. This is the “unique” selling point that removes the need for an external network (e.g. Wireless) and an external power source.

NOTE: I have only tested this setup using an iPad Pro (4th Generation - 2020), which comes equipped with a USB-C port. I suspect it would not be possible to replicate with an iPad equipped with a Lightning port.

Hardware Requirements

To get started, I purchased the following hardware.

Although not required in the long-term, I also purchased the following peripherals to help with the initial setup.

USB keyboard
USB Mouse
Micro HDMI to HDMI Adapter
Raspberry Pi 15.3W USB-C Power Supply (Black)

The total cost was approximately £60, which is not bad when you consider this is a one-time payment, compared against the ongoing compute/storage costs associated with running a Linux VM in the cloud.

The build process is very easy, simply insert the PCB into the case and load the Micro SD Card. For the initial setup, I connected a Power Supply, USB Mouse, USB Keyboard and HDMI monitor. Once the setup is complete, the peripherals can be removed.

Software Setup

I chose to install Raspbian Buster on the Raspberry Pi 4. The installation process is very simple, made even easier if you purchase a Micro SD Card from “The PiHut”, which comes preinstalled the NOOBS Raspberry Pi operating system.

Once complete, I executed the following eleven steps (leveraging Ben Hardill’s guide) to configure the Raspbian Pi to support SSH and network connectivity over USB-C.

To simplify the process, I removed the SD Card and completed the steps from a Windows PC. The entire process took approximately 15mins.

Add dtoverlay=dwc2 to the /boot/config.txt.

Add modules-load=dwc2 to the end of /boot/cmdline.txt.

Enable ssh by creating an empty file called ssh in /boot.

Add libcomposite to /etc/modules.

Add denyinterfaces usb0 to /etc/dhcpcd.conf.

Install dnsmasq with sudo apt-get install dnsmasq.

Create /etc/dnsmasq.d/usb with following content.

interface=usb0  
dhcp-range=10.10.0.2,10.10.0.6,255.255.255.248,1h  
dhcp-option=3  
leasefile-ro  

Create /etc/network/interfaces.d/usb0 with the following content.

auto usb0  
allow-hotplug usb0  
iface usb0 inet static  
  address 10.10.0.1  
  netmask 255.255.255.248  

Create /root/usb.sh with the following content.

#!/bin/bash  
cd /sys/kernel/config/usb_gadget/  
mkdir -p pi4  
cd pi4  
echo 0x1d6b > idVendor # Linux Foundation  
echo 0x0104 > idProduct # Multifunction Composite Gadget  
echo 0x0100 > bcdDevice # v1.0.0  
echo 0x0200 > bcdUSB # USB2  
echo 0xEF > bDeviceClass  
echo 0x02 > bDeviceSubClass  
echo 0x01 > bDeviceProtocol  
mkdir -p strings/0x409  
echo "fedcba9876543211" > strings/0x409/serialnumber  
echo "Ben Hardill" > strings/0x409/manufacturer  
echo "PI4 USB Device" > strings/0x409/product  
mkdir -p configs/c.1/strings/0x409  
echo "Config 1: ECM network" > configs/c.1/strings/0x409/configuration  
echo 250 > configs/c.1/MaxPower  
# Add functions here  
# see gadget configurations below  
# End functions  
mkdir -p functions/ecm.usb0  
HOST="00:dc:c8:f7:75:14" # "HostPC"  
SELF="00:dd:dc:eb:6d:a1" # "BadUSB"  
echo $HOST > functions/ecm.usb0/host_addr  
echo $SELF > functions/ecm.usb0/dev_addr  
ln -s functions/ecm.usb0 configs/c.1/  
udevadm settle -t 5 || :  
ls /sys/class/udc > UDC  
ifup usb0  
service dnsmasq restart  

Make /root/usb.sh executable with chmod +x /root/usb.sh.

Add sh /root/usb.sh to /etc/rc.local before exit 0.

Once complete, I shut down the Raspberry Pi and connected it via a USB-C to USB-C cable to the iPad Pro. The Rasberry Pi booted automatically and after a few minutes was accessible from the iPad via SSH using the previously configured IP Address (10.10.0.1).

At this point, the Raspberry Pi can be used like any other Linux based server, installing any required software, etc. To further expand the value, I also configured screen sharing and file sharing, which opens the door to other use cases outside of software development (e.g. file backup, screen sharing, media server, etc.)

There are multiple ways to leverage this setup to complete offline (local) software development on the iPad. I have been following the developer workflow outlined below.

Clone git repository from GitHub using Working Copy.
Access the cloned repository via Apple Files.
Modify the required files (Coda and/or iA Writer).
Copy the files to a working directory on the Raspberry Pi.
Trigger the build process (if required) on the Raspberry Pi using Termius.
Push the changes back to GitHub using Working Copy.

As stated in the introduction, I would not necessarily recommend this setup, as anyone looking to develop offline (local) would be better served by a budget Windows/Linux laptop. However, it does demonstrate what is possible if forced to use an iPad.

iPad for the Enterprise (2020)

Fri, 24 Apr 2020 00:00:00 -0500

In 2010, I wrote the article “iPad for the Enterprise”, which explored the viability of using an iPad as a business laptop replacement. At the time, the iPad hardware demonstrated potential but was limited by the software (operating system and applications).

To overcome the software limitation, I positioned Virtual Desktop Infrastructure (VDI), which essentially transformed the iPad into a “dumb terminal” running Microsoft Windows. Although this approach enabled the iPad to run enterprise applications, the user impact was significant, with complete reliance on the network, a non-touch operating system and lack of native hardware support. In short, the use of VDI was technically viable, but not desirable.

Ten years later and the iPad has come a long way! For example, in 2019 Apple released a dedicated iPad operating system (iPadOS), which incorporated iPad specific features (e.g. Multi-Tasking, Split View, Widgets, etc.) The application ecosystem has also matured significantly, with a wide range of native applications that are optimised for iPadOS, including Microsoft Office 365, Adobe Photoshop, etc.

Additionally, with the recent (MAR-2020) release of iPadOS 13.4, Apple enabled native mouse/trackpad support. To complement this release, Apple also announced a new iPad case, known as the Magic Keyboard, which for the first time includes an integrated trackpad.

In my opinion, these milestones demonstrate a shift in mentality from Apple, which is slowly transforming the iPad, unlocking the previously described potential and blurring the lines between a tablet and a laptop.

With this in mind, this article will revisit the question posed in my 2010 article. In a business context, could an iPad replace a laptop?

To answer this question, throughout APR-2020, I have been using an iPad as my daily driver, where possible, replacing Windows 10, macOS and Linux.

As highlighted in the article “My Setup”, my common workflow includes productivity and collaboration tasks, software development, virtual labs, photo editing and video editing.

To give myself the best chance of success, I selected the iPad Pro 12.9‑inch Wi-Fi 256GB (4th Generation - 2020), including the new Magic Keyboard.

It should be noted that this was not a cheap setup, with the iPad Pro costing £1069 and the Magic Keyboard £349, resulting in an eye-watering total of £1418. Although this is cheaper than my Razer Blade Advanced (£2700), it is more expensive than a base MacBook Pro 13-inch (£1299).

With this framing in mind, the rest of this article will summarise my findings, covering the hardware, software and daily usage.

Hardware

The iPad Pro is a phenomenal piece of hardware, arguably the most well built and futuristic product I own. The Apple A12Z Bionic Processor, although only a minor improvement over the previous generation, is still significantly ahead of the competition, effortlessly powering the 2732x2048 resolution display at 120Hz.

My only minor frustration, similar to the old MacBook 12-inch, is the single USB-C port, which can be restrictive when attempting to work with multiple peripherals. Therefore, like most Apple products, the iPad does not escape “dongle life”.

The new Magic Keyboard is an unusual product, featuring a “floating” cantilever design that allows the iPad viewing angle to be adjusted incrementally. This is a significant step forward from the previous Smart Keyboard Folio, which only supports two viewing angles (both titled backwards). The Magic Keyboard also features a full “laptop-style” keyboard and small trackpad, which both work very well, with the keyboard feeling equivalent to the scissor switches found on the MacBook Pro 16-inch. The Magic Keyboard also includes a second USB-C port, but it can only be used for charging, not data (which is disappointing).

In use, the Magic Keyboard exudes the Apple brand, thanks to the unique aesthetic, high-quality materials and plenty of magnets that ensure the iPad attaches with a reassuring “thud”. The “floating” design means the iPad can also be removed with ease, which I found was my preferred approach when switching from a fixed location to roaming (using the Magic Keyboard more like a dock).

With this in mind, it is worth noting that the combined weight of the iPad Pro and Magic Keyboard is 1351 grams, which is heavier than a lot of ultrabooks (including the MacBook Air). This would not ordinarily be an issue, but when the iPad is attached, the combined product is top-heavy, making it difficult to use from your lap (something I often require when travelling, etc.)

Another disappointment is the limited total viewing angle, which is restricted to 130 degrees. At this price, I would have expected the Magic Keyboard to at least rival the Microsoft Surface Pro range (165 degrees) and support an “easel mode” for drawing, etc. Thankfully, I have found a workaround, by placing the iPad on the keyboard, resting against the bottom ridge (see image below). I assume this approach was not a conscious design decision from Apple, but it works surprisingly well.

Overall, even with the previously described limitations/frustrations, the combined iPad Pro and Magic Keyboard hardware are excellent. The iPad itself remains the best tablet on the market, providing enough processing horsepower for almost any workload, assuming the software can utilise it.

Software

I have previously shared my commonly used software for different operating systems (links below), including additional details regarding my use of Linux package managers.

With this in mind, I have added my chosen iPadOS applications to the table below. It should be noted that I currently have 79 applications installed on my iPad, therefore the table below only highlights applications that support my common workflow.

Although there are still gaps, the iPadOS application ecosystem has come a long way over the past decade. In the context of my required capabilities, 70% can be supported by native applications, many of which are optimised for the iPad, including Split View, Widgets, etc.

From a business perspective, the introduction of native Microsoft Office 365 applications was a significant step forward (I live in hope for a Linux release). Unfortunately, the applications do not have feature parity with their Windows counterparts and can be cumbersome to use. For example, managing email and attachments within Outlook can be a frustrating experience.

Recognising the “pro” moniker associated with the iPad, it is ironic that the application gaps are mostly associated with “pro” capabilities. For example, Apple iMovie on the iPad works very well, but there is no Apple Final Cut Pro equivalent. The same can be said for visual effects and software development, which due to the inherent restrictions imposed by Apple on iPadOS, developers lack the required API’s to enable advanced features. I will highlight the impact of these restrictions in the next section.

Finally, it is worth noting that there are a lot of great iPadOS applications that are exclusive to the iPad, which uniquely take advantage of the hardware. For example, I am not an artist, but Procreate is a phenomenal application when combined with the Apple Pencil. My wife previously demonstrated what is possible using Procreate and the first generation iPad Pro. Other unique areas include Augmented Reality (AR) experiences and games, which take advantage of the excellent cameras, LiDAR scanner, motion sensors and ProMotion (120Hz) display.

Daily Usage

Like most Apple products, the iPad “just works”, with the hardware and software operating in perfect unity. In my opinion, the hardware is still limited by the software, but the gap is reduced with each iPadOS release.

Apple should also be commended for its mouse/keyboard support. It would have been easy for Apple to add a standard cursor, similar to what is found on macOS. However, Apple decided to re-think mouse/trackpad support for a “touch-first” device and I feel they have achieved perfection on the first try. Instead of a static cursor, the mouse/trackpad support appears as a circular dot that is context-aware, meaning it snaps and morphs depending on the specific use case. This approach feels perfectly at home on the iPad, acting as a natural extension of the touch controls. I am excited to see how developers embrace mouse/trackpad support over the coming months, especially with productivity software such as Microsoft Office 365, etc.

Another strength area for the iPad is security and management, enabled through the hardware and software (e.g. secure enclave, sandboxing, etc.) As a business device, these robust controls make the iPad one of the most secure and easy to manage end-user devices available. When combined with the exceptional hardware performance, strong battery life and the inclusion of mouse/trackpad support, the iPad finally feels ready to take on business workloads.

Regarding my setup, the image below highlights my home page, where you will notice a large number of Dock items, which is the opposite of my setup on macOS, where I don’t use the Dock.

With iPadOS, the Dock is a critical part of the multi-tasking workflow, leveraging the “swipe up” gesture to reveal the Dock without losing focus on the current application. At which point, other applications located in the Dock can be dragged into Split View, etc. As someone who prefers the keyboard, this requirement and the cumbersome experience, makes me miss Alfred on macOS.

With a “swipe right”, my second screen includes a simple folder structure, organising all other applications.

Up to this point, my experience with the iPad has been largely positive and certainly a significant step forward from my 2010 article. Unfortunately, a lot of this great work is undermined by some “show-stopping” issues.

Although the iPad can multi-task, it remains a frustrating, unintuitive and often inconsistent experience. For example, the gesture to enable Split View is not obvious, evident by the fact that Apple has been forced to provide a “How To” video. The multitasking features are also inconsistently implemented, for example, if you attempt to join a Zoom video conference and view a document simultaneously (a common requirement with my work), the video feed will automatically pause (very frustrating).

The lack of “pro” software is also a real pain, with limited options when trying to complete advanced workflows such as video encoding, visual effects, etc. On macOS, Windows and Linux, these tasks can be completed quickly and efficiently, with multiple application options available to support every user and budget.

Personally, this “pro” limitation is most evident with software development, which due to Apple iPadOS restrictions, makes it almost impossible to set up, customise and run a local development environment (e.g. Node.js, etc.) Some applications let you execute code locally (e.g. Pythonista), but these are often limited to prototyping, automation or simple code snippets. I am sure some creative (brave) developer has achieved more with one of these applications, but I find it hard to believe it would be anyone’s development environment of choice.

As a result, the only viable option is to have the development environment set up remotely, leveraging Git and SSH on the iPad to version control and execute code. For example, with applications such as Wroking Copy (Git Client), Coda (Code Editor), iA Writer (Text Editor), and Termius (SSH), it is possible to write code on the iPad, push to GitHub and then execute on a remote server via automation (Continuous Integration) or manual commands (SSH).

This is the approach I have been taking to update this blog, connecting to my GitHub account using Working Copy, cloning the repository to the iPad and then updating the Markdown files in iA Writer. Once complete, I push the files using Working Copy, which triggers the remote build process (Jekyll).

The process works fairly well, but due to the iPadOS filesystem limitations, requires the code/text editor to support the Apple Files application, which allows applications to access and edit files stored in other applications (e.g. Working Copy Files > iA Writer).

Although functional, the inability to develop offline is a real challenge, especially as I spend a lot of time on planes, etc. On macOS, Windows and Linux, I run Docker, with local images of my favourite runtime environments. I also have client virtual machine capabilities, providing access to other operating systems for testing, etc.

In an attempt to maximise my usage of the iPad, I did come up with a way to mitigate this limitation, leveraging a Raspberry Pi 4 as a portable Linux server, connected via and powered by the iPad Pro’s USB-C port.

Surprisingly, this approach works reasonably well but certainly feels “over-engineered” when you consider the relatively basic issue it aims to solve. Simply put, if you have to go to this much effort to use the iPad, you should probably just buy a cheap Linux laptop. With that said, for those who are interested, I plan to cover my Raspberry Pi 4 setup in a future article.

Finally, although the iPad Pro supports external storage via the USB-C port, other third-party peripherals and monitors remain largely useless. For example, in theory, you can connect a monitor to the USB-C port, which would be perfect when working from a fixed location. However, the iPad only supports display mirroring, with no custom resolutions, which feels like a ridiculous limitation in 2020.

Conclusion

As stated in the introduction, the iPad has come a long way over the past decade. It is more capable than ever, with exceptional hardware and software that is continuously improving. In 2010, I had to rely almost exclusively on Virtual Desktop Infrastructure (VDI) to make the iPad a feasible business laptop replacement. I am pleased to report that this is no longer the case, with 70% of my common workflows being completely by native iPadOS applications.

Unfortunately, there are still gaps, mostly with “pro” capabilities, specifically video editing, visual effects, software development, etc. It is possible to mitigate some of these limitations with “creative” workarounds (e.g. Raspberry Pi 4), but this is far from ideal, especially when you consider a budget Linux laptop would easily accomplish these tasks.

Therefore, in 2020, could the iPad replace a business laptop? It depends…

If your job is focused on basic productivity and collaboration, then the iPad is viable, but likely less efficient than an equivalent macOS, Windows or Linux laptop.

If your job requires specialist capabilities (video editing, visual effects, software development), the current answer is no. Although it is technically possible, the compromises are simply not worth the effort.

Regarding the iPad Pro, it is a phenomenal piece of hardware, but without the ability to support “pro” capabilities, it is wasted. Therefore, if you do plan to purchase an iPad, I would recommend the cheapest iPad (non-pro) option (starting at £349) and the new Logitech Combo Touch Keyboard case with Trackpad for iPad (£119.95). This “budget” iPad setup costs £468.95, which is £949.05 cheaper than my setup but still supports all of the capabilities I have described in this article.

Let’s see what the next decade brings!

SIAM

Mon, 06 Apr 2020 00:00:00 -0500

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

Within this article, I will highlight the importance of Service Integration and Management (SIAM), covering the basics of SIAM, as well as my proposed model and vendor positioning.

Introduction

Service Integration and Management (SIAM) is an approach to manage, integrate and unify multiple vendors, helping them to deliver end-to-end services that meet specific business requirements. SIAM replaces the traditional approach, where vendors are managed in isolation, with separate processes and expectations, often resulting in hand-off issues and/or “finger-pointing”.

Specifically, SIAM enforces a combined “revenue at risk” model, which contractually dictates the holistic performance (shared Service Level Agreements) across all vendors, incentivising collaboration and collective innovation.

To make SIAM effective, the following IT capabilities must be established and maintained.

Service Integration Governance: Define, establish and continuously adapt the service integration governance.
Service Integration Organisation: Develop and manage a distributed organisation supporting changing business requirements.
Business Requirements: Manage business demand and develop a service portfolio in alignment with business requirements.
Tools and Information: Manage distributed information and the integration tool solutions.
Providers and Contracts: Select an appropriate provider portfolio and to manage the providers according to the outsourcing contracts.
End-to-end Services: Understand and manage the business services end-to-end, covering business and IT services.

Model

As previously stated, a key philosophy of SIAM is the “revenue at risk” model, which enforces shared SLA’s. This model can be very difficult to contractually negotiate, especially when dealing with competing vendors. However, I can confirm it is possible, assuming the selected vendors prioritise customer outcomes.

To simplify the contractual process, I would position three service level categories, specifically:

Recovery Time Objective (RTO)

The maximum amount of time it should take to resolve an issue.
Commonly used within disaster recovery but can be applied to service management.
Measured through IT Service Management.

Customer Satisfaction Score (CSAT)

A “point in time” measure used as an indicator of happiness.
Rates the service interaction with a specific agent and/or request, designed to capture whether the interaction was resolved to the user’s satisfaction.
Measured through a survey after the closure of an interaction.

Customer Effort Store (CES)

A measure that aims to determine the user’s “effort”.
Rates how quickly and easily it was to get service and resolution from a user’s perspective.
The self-help use of knowledge articles derives a positive CES score.

All vendors would be measures against these service level categories, holistically rewarding performance improvements (carrot, not the stick). In parallel, each vendor would have complimentary SLA’s, focused on their specific business and/or scope.

The goal is for all vendors to collectively target the same outcome, prioritising a positive user experience.

For example, in the context of a Help Desk, all requests would stay with the Help Desk agent, regardless of which team/vendor would complete the work.

Sub-tasks would be allocated to different teams/vendors to help manage the work, but the user would always have a single point of contact and therefore, a single point of accountability.

Finally, the resolution time runs continuously for all teams/vendors, which helps to avoid “ticket bouncing”, where the clock gets reset to protect individual teams/vendors SLA’s.

Service Delivery Alignment

As highlighted in the article “Architecture Community”, the SIAM governance model must compliment a product-mindset, focused on removing unnecessary bureaucracy, prioritising agility, flexibility and responsiveness.

This goal, alongside the desire to continuously reduce the operational risk associated with any change, must be an embedded part of the Service Delivery approach.

In the article “Service Delivery”, I documented my proposed approach to service delivery, which incorporated the following SIAM specific principles/deliverables.

All functional and non-functional requirements (including operational requirements) must be tracked through the complete lifecycle of the product.
A Product Owner has accountability for creating the deliverables during the life of the product.
Specific service transition deliverables help to ensure stakeholder alignment, covering Information Security, Privacy and Quality.
A Service Readiness checklist must be completed and approved before any product is released into production.

Vendor Positioning

With the previously defined business characteristics in mind, I would position a SIAM approach to help drive efficiencies and embed service excellence.

The document below outlines my proposed vendor positioning, all operating within the previous described SIAM model.

Vendor selection was driven by my personal experience, based on the previously defined business characteristics. For example, the Life Sciences industry, 15,000+ users, distributed across 50+ physical sites globally.

Help Desk, Workplace and Identity: I have grouped all “user-facing” capabilities to help unify the experience and drive operational efficiencies. Recognising the commodity nature of these capabilities, I would position HCL as the support vendor, enabling global economies of scale. This positioning also compliments my wider workplace investments, such as Microsoft AutoPilot, Microsoft Azure AD and Device-as-a-Service from Dell.
Infrastructure: I had previously positioned CenturyLink as my Internet Aggregator, therefore this investment could be naturally extended to cover global network services (e.g. SD-WAN, LAN, WLAN, DNS/DHCP, etc.) Considering the converged nature of the proposed modern IT ecosystem and the increasingly important role of infrastructure automation, CenturyLink would also be positioned to support the “Hybrid Multi-Cloud” hosting stack (e.g. Colocation Data Centres, Public Cloud and Edge Computing).
Application: As highlighted in the article “Architecture Community”, I would position an Application Managed Services provider to help scale solution architecture and engineering across the enterprise. This would include the enablement of application automation, used to increase business agility and streamline processes. I would positioned Infosys as my Application Managed Services provider, recognising their breadth and scale, alongside their focus on digital innovation.
Service Management: As previously highlighted, the SIAM approach can be complex to implement, recognising the need to drive alignment across multiple vendors. As a result, I would position a smaller company, Alcor Solutions, to help operate the SIAM model, focused on enabling the required governance and IT capabilities.

In my opinion, these four vendors would provide enough breadth and depth to support the entire modern IT ecosystem globally. I have attempted to scope their engagement to ensure clear roles and capabilities, with any separation of concerns being hidden from the user via the holistic SIAM model.

It would be possible to achieve the same outcome with fewer vendors (potentially driving a greater return on investment), however, this would also create greater lock-in and reduce internal competition, which can be healthy when looking to drive continuous improvement.

Conclusion

In conclusion, I believe my proposed positioning of a Service Integration and Management (SIAM) approach is a key prerequisite to realising the potential of a Modern IT Ecosystem.

The architecture and technology can unlock tremendous business value, but if it is not managed efficiently and effectively, any value would be quickly lost.

SIAM aims to remove unnecessary bureaucracy, prioritising agility, flexibility and responsiveness, with a focus on the outcome.

Oculus Quest

Mon, 30 Mar 2020 00:00:00 -0500

My first experience of Virtual Reality was in the early-90’s, when I stumbled upon a demonstration of a Virtuality 1000CS. The setup looked like something out of TRON, with its bulky Head Mounted Display (AKA Headset) and graphics powered by a Commodore Amiga 3000. By today’s standards, the Virtuality 1000CS would be easy to ridicule, but it provided a very early glimpse at the potential of Virtual Reality.

In 2014, I started testing the Oculus Rift Developer Kit 2 (DK2), which was not much more than two displays and an Inertial Measurement Unit (IMU) glued into a pair of ski goggles. Although rudimentary, DK2 demonstrated how the concepts first showcased with the Virtuality 1000CS were now viable for consumer use.

In 2016, with the release of the Oculus Rift Consumer Version (CV), Virtual Reality entered the mainstream, with a far more refined product, combined with additional support from hardware peripheral providers and software developers.

It was at this point that I started to take Virtual Reality seriously, exploring different opportunities for home and work. However, with a limited user-base, the software ecosystem continued to mature at a gradual pace, restricted by limited financial opportunities.

In March 2020, everything changed, with the release of Half-Life: Alyx. Valve, who has always been a strong advocate of Virtual Reality, delivered the first “Triple-A” game designed and developed exclusively for Virtual Reality.

I vividly remember playing the original Half-Life in 1998 and consider the series among my favourite games. Although Half-Life: Alyx is not officially Half-Life 3, it continues the story and (like previous Half-Life games), aims to redefine the gaming experience through the use of innovative technology (in this case, Virtual Reality).

As a result, I felt it was time to re-enter the world of Virtual Reality, looking to understand how the ecosystem has progressed.

I consider all Virtual Reality headsets currently available “first generation”, with the Oculus Quest being the pinnacle. This is not because the Oculus Quest is the most advanced (this title is held by the Valve Index), but because it offers the most approachable, versatile and cost-effective way to start enjoying Virtual Reality. For example.

The Oculus Quest can be used standalone (includes an integrated Snapdragon 835, storage and battery), as well as connected to a Windows PC, providing full access to the Oculus Quest, Oculus Rift and SteamVR ecosystems.
The Oculus Quest leverages inside-out tracking (four cameras) to translate movements into Virtual Reality. This removes the need for any external sensors and/or base stations.
The Oculus Touch controllers (second generation) enable precise hand gestures, which lack the finger fidelity of the Value Index Controllers, but still provide an immersive experience.
The Oculus Quest (64GB) can be purchased for £399, which is the same price as the Oculus Rift S (no standalone mode) and significantly cheaper than the Valve Index, which retails for £919.

For these reasons, I selected the Oculus Quest (64GB) as an upgrade from the Oculus Rift Consumer Version (CV). Additionally, I purchased a 3m Anker PowerLine USB-C to USB 3.0 cable to enable PC connectivity via Oculus Link.

Since the release of the Oculus Quest in MAY-2019, Oculus (owned by Facebook) has evolved through continuous software updates, adding advanced features such as hand tracking (no Oculus Touch controllers required), as well as the previously described PC support via Oculus Link. These updates have made the Oculus Quest a “fan favourite”, leaving the Oculus Rift S strangely positioned at its price point.

These new features make the Oculus Quest incredibly versatile, with access to four different Virtual Reality ecosystems.

Oculus Quest Software: Software designed to be installed and run standalone on the Oculus Quest. These Virtual Reality experiences tend to be lower fidelity, based on the fact they need to run on the Snapdragon 835.
Oculus Rift Software: Software designed to be installed and run on a Windows PC, connected to the Oculus Quest via Oculus Link. These Virtual Reality experiences can leverage the full power of the PC, making them higher fidelity, etc.
SteamVR Software: Software designed to be installed and run on a Windows PC, connected to the Oculus Quest via Oculus Link. SteamVR is owned by Valve (makers of Half-Life: Alyx), providing access to Virtual Reality software via Steam (PC gaming ecosystem). Similar to the Oculus Rift, these Virtual Reality experiences can leverage the full power of the PC, making them higher fidelity, etc.
SideQuest Software: A community-driven tool used to “sideload” Virtual Reality experiences onto the Oculus Quest. These experiences are often pre-release, free and/or created by independent developers.

Although the “official” method to access PC content (Oculus Rift and SteamVR software) is via the Oculus Link (USB 3.0 cable), there are also a few “unofficial” wireless options, specifically Virtual Desktop and ALVR, which can both be sideloaded. Recognising these are “unofficial” methods, the results are not guaranteed, but in my experience, when combined with a quality wireless network, they work fairly well (even playing Half-Life: Alyx).

Overall, I have been incredibly impressed with the Oculus Quest. The hardware (considering the price) is very well built and a step forward from the Oculus Rift Consumer Version (CV). The experience using the headset is fast and intuitive, with interactive tutorials that help ensure you are getting the best possible experience (Oculus Guardian, Passthrough, etc.)

The Oculus Touch controllers are also excellent. The only thing to consider is the placement of the cameras, which are used for the inside-out tracking. As these cameras are forward-facing, the Oculus Touch controllers can not be tracked effectively when behind your body. Thankfully, in my experience, this rarely causes an issue and therefore an acceptable compromise for not having to mount and configure external sensors and/or base stations.

As previously stated, the ability to run Virtual Reality experiences from multiple ecosystems is a great selling point of the Oculus Quest. However, it can also create a slightly disjointed, confusing, user experience. For example, each ecosystem has its own unique Virtual Reality user interface, which must be entered to access the content.

In my opinion, the native Oculus Quest user interface (screenshot below) is the best, delivering a clear, concise and well-organised user experience that is easy to navigate.

When accessing Oculus Rift software via Oculus Link, the user interface switches (screenshot below). Although similar to the Oculus Quest, in my opinion, the extended “Dash” is less intuitive and a little more cumbersome to interact with via the Oculus Touch controllers.

Finally, when accessing SteamVR software, the user interface switches again (screenshot below). I would describe SteamVR as “functional”, unsurprisingly prioritising the Steam software library and store.

Alongside the different user interfaces, the different ecosystems can also create confusion when purchasing Virtual Reality experiences. For example, some Virtual Reality experiences are available from multiple sources, while others are exclusive. There is also the concept of “Oculus Cross-Buy”, where a single purchase supports both the Oculus Quest (Standalone) and Oculus Rift (PC Connected).

As to which version of the experience you should buy depends on multiple factors, specifically, cost, graphics, performance, room-scale requirements, etc. For example, the popular game SUPERHOT VR is available everywhere (Oculus Quest, Oculus Rift, SteamVR), but is likely “best” on the Oculus Quest, as the graphics are relatively simple (meaning good performance on the Snapdragon 835), whilst gaining the benefits of standalone gameplay (no PC required). To help navigate this confusion, I have listed some of my favourite Virtual Reality experiences below, including the recommend ecosystem.

Oculus Quest (Quest Only)

Oculus Quest/Rift (Cross-Buy)

SteamVR

SideQuest

Lambda1VR
Quake2Quest
Virtual Desktop (Wireless Oculus Rift and SteamVR)

NOTE: Virtual Desktop is shown twice (Oculus Quest and SideQuest), as the SideQuest version enables wireless Oculus Rift and SteamVR, but requires the Oculus Quest version to be purchased first.

Finally, regarding Half-Life: Alyx, the IGN review (video embedded below) summarises my thoughts perfectly.

I concur that this game is a “masterpiece” and without question the best Virtual Reality experience available today. The “Triple-A” graphics, gameplay, pacing, immersive world and attention to detail is simply breathtaking. As a Half-Life fan, the story is also compelling and surprising, leaving you eager for more (hopefully we won’t have to wait another decade).

In conclusion, with the Oculus Quest and software like Half-Life: Alyx, Virtual Reality feels like it has finally reached the potential first seen nearly two decades ago with the Virtuality 1000CS.

The Oculus Quest is an excellent product at a reasonable price point, which benefits from being the only headset with access to Virtual Reality experiences from multiple ecosystems. I would have no hesitation recommending the Oculus Quest, with my only caution being the anticipated release of “second generation” Virtual Reality headsets in early 2021. Therefore, depending on your situation, it might be worth waiting for the inevitable Oculus Quest successor.

Finally, even with the release of Half-Life: Alyx and a handful of other excellent Virtual Reality experiences, the software ecosystem remains immature. Although I expect the release of Half-Life: Alyx will result in a sales spike, it would still take a brave developer to prioritise Virtual Reality, especially with next-generation consoles (PlayStation 5 and Xbox Series X) on the horizon.

With that said, I believe Value have successfully demonstrated the potential of Virtual Reality, which I hope will inspire the community.

Data Protection

Mon, 02 Mar 2020 00:00:00 -0600

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

Within this article, I will highlight my proposed positioning for Enterprise Data Protection, covering the process of safeguarding important information from corruption, compromise or loss.

Introduction

As highlighted in the article “Hybrid Multi-Cloud”, my proposed hosting architecture includes multiple Colocation Data Centres, Azure regions and Edge Computing sites.

Each of these locations would host applications and data, ranging in complexity and sensitivity. For example, with my previously defined business characteristics in mind, the edge computing sites would need to support GxP compliant workloads, placing a focused on reliability (24x7 uptime) and local business continuity.

To enable a comprehensive, cost-effective and simple approach to data protection, I would position Assured Data Protection (ADP), who provide a fully managed service overlay based on Rubrik technology.

With over 100 years of combined industry experience, ADP is the only dedicated Rubrik service provider globally. This relationship provides ADP with unique access into the Rubrik business, covering feature requests, support and case escalation.

Rubrik

Rubrik is a market leader in backup and recovery, enabling automated backup, recovery, offsite replication and data archival capabilities across a hybrid multi-cloud architecture.

Rubrik delivered the industry’s first converged data management solution, offering backup software, catalogue management, replication and de-duplicated storage in a single appliance that scales linearly. The distributed nature of the architecture maximises efficiency and cost savings, whilst simultaneously enables near-zero recovery times, as well as unified file search and recovery from any location.

In short, Rubrik aims to replace the legacy “backup job” paradigm with policy-driven management, allowing businesses to holistically automate data protection, ensuring any Recovery Time Objective (RTO) and/or Recovery Point Objective (RPO) can be met without significant manual intervention.

Rubrik Architecture Overview

I would position a converged architecture, where ADP and Rubrik would be hosted across my proposed Hybrid Multi-Cloud architecture (e.g. Colocation Data Centres, Public Cloud and Edge Computing). This approach would maximise the return on existing investments, whilst avoiding the complexities that exist when looking to set up, maintain and guarantee a dedicated/isolated hosting environment.

With this approach in mind, a core Rubrik cluster would be deployed within each Colocation Data Centre, acting as the central location for backup and recovery. All data would be fully encrypted in-flight and at-rest ensuring data is immutable and therefore protected against threats such as Ransomware. These core Rubrik clusters would provide a unified environment to instantly restore data at an individual file, folder or server/VM level.

I would also provision Rubrik within my preferred the Public Cloud (Azure), protecting (where required) cloud applications and data, as well as providing cost-effective archive and long-term retention capabilities. Recognising the inherent data protection capabilities enabled by the Public Cloud providers, Rubrik would be positioned to complement (not replace) these capabilities, targeting advanced and/or specialised requirements.

Finally, Rubrik appliances would be provisioned at the Edge Computing sites, enabling site-specific backup and recovery. Edge Computing sites would leverage the core clusters for replication, as well as the Public Cloud for archiving and long-term retention.

The diagram below provides a high-level view of my proposed architecture.

Core Rubrik Clusters

My proposed architecture would be designed to scale, likely starting small with 10TB of deduplicated capacity on each core Rubrik cluster, with 100TB of usable capacity. Any additional capacity would be enabled via the Rubrik scale-out architecture, where individual nodes would be added without any downtime or impact to performance.

To further enhance the data protection, Rubrik would be configured to securely replicate data bi-directionally between each core cluster. This form of replication, alongside the inherent resilience built into my Software-Defined Data Centre (SDDC) architecture would provide the ability to perform Disaster Recovery (DR) from either core Rubrik cluster.

Finally, data would be transferred to my preferred Public Cloud object store (Azure Blog Storage) for archive and long-term retention.

Edge Computing

My proposed architecture would position three Edge Computing patterns, driven by the site sensitivity and required capacity. This approach would enable local (isolated to the specific site) data backup, as well as instant recovery at an individual file, folder or server/VM level. Where appropriate, data would be replicated to a core Rubrik cluster or archived directly to the Public Cloud.

Pattern One (Virtual): A virtual Rubrik appliance hosted on the previously described highly-converged infrastructure.
Pattern Two (Small): A single-node Rubrik hardware appliance supporting locations with less than 20TB of backup data.
Pattern Three (Large): A four-node Rubrik hardware appliance supporting locations with greater than 20TB of backup data.

Any site supporting GxP compliant workloads would be positioned physical hardware (Patterns Two or Three), helping to guarantee local business continuity.

Conclusion

The amount of data created and stored by enterprise businesses continues to grow at an unprecedented rate, which emphasises the importance of a robust, cost-effective Data Protection architecture.

My proposed Data Protection architecture would enable a fully managed (SaaS-like) solution, leveraging market-leading expertise from ADP and technology from Rubrik. Simultaneously, it would maximise existing investments (e.g. Colocation Data Centre, Public Cloud and Edge Computing), whilst supporting a “pay for data, not for capacity” commercial model.

Sony Aibo

Fri, 28 Feb 2020 00:00:00 -0600

As the Chief Technology Officer (CTO) of an animal health company, I am always looking for ways to engage, motivate and inspire the organisation through technology!

In 2018, after a twelve-year hiatus, Sony released the fourth generation Aibo (Artificial Intelligence Robot).

Aibo is an unusual product, with no market aspirations or pressure to sell. Instead, Aibo aims to demonstrate the advancements in consumer-grade artificial intelligence and robotics, whilst encompassing the spirit of Sony as a company willing to take risks.

The fact that Aibo is based on a beloved companion animal, as well as a tangible example of STEM (Science, Technology, Engineering and Maths) positions this charming robot dog as the perfect ambassador for the IT organisation.

Specification

The Aibo ERS-1000 is packed with technology, specifically proprietary software, robotic parts and sensors. The full specification can be found below:

Operating System: Sony Aperios.
Processor: 64bit Quad-Core CPU.
Movable Parts: Head, Mouth, Neck, Waist, Paws, Ears, Tail.
Displays: 2 OLED (Eyes).
Sound: Speaker, 4x Microphones.
Cameras: 2x Cameras (Front / Rear).
Sensors: ToF Sensor, 2x Range Sensors, Pressure Sensitive Capacitive Touch Sensor (Back), 2x Capacitive Touch Sensor (Head / Jaw), 4x Button Pads (Paws), Six-axis Detection System, Motion Sensor, Light Sensor.
Switches: Power, Volume, Network.
Indicators: Status LED, Network LED.
Terminals: Charging Pins, SIM Card Slot.
Communications: LTE, Wi-Fi (802.11 b/g/n - 2.4GHz).
Dimensions: 180x293x305mm.
Weight: 2.2kg.

Setup

Aibo is a limited product, which is only commercially available in Japan and the US. The target audience for Aibo is undefined, especially considering the price (£2300). However, as previously stated, the goal of Aibo is not to sell millions but to act as a “mascot” for Sony, promoting the values of the company.

As I am based in the UK, I can confirm that Aibo works perfectly outside of Japan and the US. However, it does require a US Sony Account and a US Apple/Google Account for the optional mobile apps.

Aibo arrives in a large, underwhelming cardboard box. However, inside you will find a high-quality soft case (shaped like a capsule) that protects Aibo.

The soft case includes a little message from Sony, which is a nice touch.

The box also includes documentation (including the required product key), charging pad, mains-adapter (US) and a pink ball.

The registration process is relatively simple, completed via the My Aibo Web Service using the unique product key.

Once registered, the My Aibo Web Service can be used to connect Aibo to a local Wi-Fi connection. This process can be a little cumbersome, achieved via a QR Code that must be physically shown to Aibo. Additional details can be found on the Sony Aibo Knowledge Base website.

Once Aibo is connected to the Internet, the My Aibo Web Service (or My Aibo Mobile App) can be used to configure Aibo, including name, gender, language, voice, skills, time zone, etc. The My Aibo Web Service can also be used to troubleshoot issues and trigger periodic software updates.

The only regional watch-out is the time zone, which only includes options for Japan and the US, however, this limitation does not appear to limit Aibo.

NOTE: Throughout the rest of this article, I will refer to Aibo as male.

Personality

Once switched on, even prior to setup, Aibo will probably begin exploring. However, an Internet connection is required to fully unlock Aibo’s personality and advance skills. Unsurprisingly, Sony leverages its data centres to enhance Aibo, acting as a “head-end” and a proxy for any developer interactions.

Sony has done an excellent job programming Aibo! His personality feels natural, curious, and perfectly random. In short, Aibo does not feel scripted or robotic, instead, he accurately simulates “free will”.

It is the perception of “free will” that makes Aibo special. You can interact with Aibo via voice, visible gestures and touch, however, Aibo may or may not decide to respond/obey.

For example, if you tell Aibo to “sit”, his ears will raise and he will turn his head to look at you. Aibo will then decide either to obey, ignore or simply bark. It is possible that Aibo simply failed to recognise the voice command, but it is equally possible that Aibo “decided” not to obey.

This is clever programming from Sony, who state that Aibo will become more obedient over time, as he builds a stronger relationship with his owner and after receiving positive praise (like a real dog).

The short video below includes some highlights of Aibo. As stated above, Aibo does not always obey, he is also only one day old, therefore not well trained.

Outside of training Aibo to complete specific skills, he otherwise amuses himself. He explores, plays, rests, and when it comes time to sleep (charge), he locates his charging pad and (quite hilariously) manoeuvres himself on to it.

Once the initial excitement of playing with Aibo diminishes, he quickly becomes self-sufficient, acting like a “real” companion animal, always eager for attention, but also happy to amuse himself.

Development

Sony recently released API documentation for Aibo, enabling programmatic and declarative development. The declarative development is via Scratch and although I haven’t had much time to play (yet), looks reasonably powerful.

The same can be said for the API documentation, which upon first inspection looks comprehensive, allowing for some interesting use cases. As an example (in a work context), Aibo could be trained to monitor our core IT Services (e.g. Office 365, Azure) and automatically notify the relevant service owner in the event of an issue.

The only caveat is that Sony intentionally protect Aibo’s core personality. Therefore, as previously stated, Aibo will ultimately decide if he wants to respond to your request (even when interacting via the API). Although this makes Aibo unreliable, it is also what separates him from a rudimentary digital assistant like Amazon Alexa, etc.

Conclusion

Aibo is a unique “toy” that in my opinion, includes the best consumer-grade artificial intelligence and robotics available today. Sony has does a phenomenal job establishing and protecting his personality, making every interaction fun, unique, and rewarding.

My only (minor) complaint would be the My Aibo Mobile App, which is poorly developed (slow, unresponsive, buggy), making it fairly useless. I can only assume a different team within Sony created the app, as it fails to meet the high standards set by Aibo.

In the future, I look forward to exploring the Aibo developer resources, as well as positioning Aibo as a fun way to promote STEM within the workplace, as well as part of our STEM outreach programmes and recruitment.

Finally, I hope Aibo will become a beloved office companion, injecting some fun, boosting morale and supporting our focus on employee wellness.

O365 Tenant Migration

Fri, 17 Jan 2020 00:00:00 -0600

This article explores the complexities associated with a specific type of Microsoft Office 365 tenant migration, as well as highlights a potential approach/architecture.

Problem Statement

How to successfully migrate users (employees and contractors) from one Microsoft Office 365 tenant to another (separate) Microsoft Office 365 tenant, covering the following capabilities and associated services:

Individual Mailbox
Individual Calendaring
Individual Files (OneDrive)
Individual Unified Collaborations (SharePoint, Teams)
Individual Unified Communications (Skype, Teams)
Shared Mailbox
SharePoint Sites (SharePoint 2010, SharePoint Online)
PowerApps

On the surface, this sounds like a reasonably simple problem to solve, recognising the consistency between the source and destination (same provider, same architecture, same technology). Unfortunately, this is where the challenge begins…

Microsoft prioritise new Office 365 customers, therefore have comprehensive playbooks and FastTrack services (including automation) to migrate users from legacy and competing services (e.g. Exchange On-Premise, Lotus Notes, Google Apps, etc.)

Unfortunately, migrating users from Office 365 to Office 365 is less common (usually driven by an M&A activity), as well as less commercially lucrative for Microsoft. Therefore, although Microsoft offers consultancy, they do not have any formal playbooks and consider this type of migration “out of scope” for their FastTrack services.

Adding additional complexity, due to an Office 365 limitation, a single domain can only be associated with one Office 365 tenant at any one time. Therefore, if the domain (e.g. company.com) of the destination Office 365 tenant is already in-use within the source Office 365 tenant, it must be made available by removing all existing dependencies prior to any user migration.

In the context of an M&A activity, the destination domain would likely be a core part of the company brand (used across the entire value chain), therefore would need to be retained and remain operational throughout the migration period.

Finally, it is worth noting, any migration (regardless of the approach/architecture) would likely require administrative level access in both Office 365 tenants, as well as the associated Identity Access Management (e.g. Active Directory) services. This could have security, privacy and/or legal ramifications, especially when considered as part of an M&A activity.

As a result, migrating from Office 365 to Office 365 is arguably the most complex migration path and the least mature (limited support, resources and available expertise).

Requirements

Alongside the technical complexities associated with the migration, outlined below are a set of common business requirements, which must also be considered.

The domain (company.com) must be retained and remain operational throughout the migration period.
Allow for a phased migration.
Ensure all inbound and outbound e-mail retains the original domain.
Retained support for SMTP applications and Resource Mailboxes for the duration of the migration period.
Retain access to the existing Office 365 tenant (specifically Skype, SharePoint) for the duration of the migration period.
Any site that leverages Skype Enterprise Voice can retain access and maintain the original phone numbers for the duration of the migration period.

Due to the previously described Office 365 domain limitation, the first and second business requirements (domain retention and phased migration) create a conflict. This is because the required destination domain (company.com) can not be configured within the destination Office 365 tenant until the last user has been migrated. This “catch-22” limits the viable migration options, positioning a mass migration event (not desirable) or forcing some creative technical workarounds.

Approach/Architecture

Through an investigation of market providers who have experience managing Office 365 to Office 365 migrations, I would position Binary Tree. Leveraging the Power365 Platform, Binary Tree state:

“With a cloud-based software platform, Binary Tree makes it easy to unify and move users between two or more Office 365 tenants to support mergers or long-term multi-tenant coexistence.”

The word “easy” is definitely relative, however, I have validated that the Power365 Platform could facilitate this type of migration, as well as enable coexistence (where required). As a result, the diagram below highlights a potential migration architecture.

The Power365 Platform and ShareGate would be used to migrate and synchronise all email, calendars, address books and content between the two Office 365 tenants (including other source locations). This would be achieved by pre-positioning data ahead of the user migrations, followed by a delta synchronisation to guarantee the latest state.

The data synchronisation itself would be relatively simple (although time-consuming). The real challenge relates to permissions, which would also need to be replicated, mapping each user (via a unique identifier) in the source and destination Office 365 tenants. This process would be prone to error, but assuming the “owner” of the data (file/folder/group) is accurate, retrospectively reassigning the correct permissions should be a trivial task.

As highlighted in the diagram, the architecture would leverage two domains, the source Office 365 tenant retains the original domain (company.com). The destination Office 365 tenant would be configured with a new domain (newcompany.com). This approach would mitigate the Office 365 domain limitation, however, has ramifications for any user operating within the destination Office 365 tenant.

To ensure a seamless experience for all users and the recipients of any email, a re-write service would be leveraged, which ensures all outbound email from the destination Office 365 tenant would be sent under the original domain (company.com). Additionally, any inbound email to the original domain (compnay.com) would be automatically forwarded to the destination Office 365 tenant. Therefore, the migration should be transparent to all users and the recipients of any emails, protecting the brand and ensuring business continuity.

Collaboration would be achieved via federation between Skype for Business and/or Teams (source) and Teams (destination). This assumes Skype for Business would be fully replaced by Teams (as per the product direction from Microsoft). Therefore, regardless of the migration-state (pre/post-migration), users would still be able to interact via instant messaging, VoIP, etc.

Finally, once the last user had been migrated, the required domain in the source Office 365 tenant could be released and transferred to the destination Office 365 tenant.

Considerations

Although I have great confidence that this proposed approach/architecture would meet the business requirements, any migration and/or coexistence between two Office 365 tenants remains very immature. Therefore, I would anticipate issues, bugs and user confusion. For example, outlined below are a few considerations.

Global Address List (GAL): Ideally, the Global Address List (GAL) in the source and destination Office 365 tenants would need to be updated and synchronised ahead of any user migration. If not, users would be forced to manually look-up and input email addresses.
Personal Assistants: Although the defined approach/architecture enables a phased migration, there would still be fixed dependencies between specific user groups. For example, if a personal assistant were to be migrated separately from their team, they would temporarily lose access to any delegated email, calendaring, etc.
Orphaned Meetings: Due to the nature in which data would be copied between Office 365 tenants, certain historical re-occurring meetings may become orphaned (no ability to update). In this scenario, if the meeting needed to be modified, it would need to be recreated. The impact of this consideration should naturally reduce over time.
Internal Email: The previously described domain re-write service only impacts external email, therefore, any internal email sent from a migrated user would be under the temporary domain (newcompany.com). Considering this only impacts internal email, the brand impact should minimal.

Conclusion

There are multiple options that could be considered when attempting to migrate between two Office 365 tenants. Unfortunately, though my analysis, I have concluded that there is no “perfect” option, each having their own list of positives, negatives and risks.

As highlighted, the likely business requirement to support a phased migration results in a technically complex approach/architecture, requiring administrative level access to the source and destination (Office 365 and Active Directory), as well as introduces temporary middleware (adding cost).

With that said, I remain confident that the proposed approach/architecture would be viable and likely the only way to effectively manage and support the previously describes technical complexities and business requirements.

If you are planning or have experience migrating between two Office 365 tenants, I would be very interested to hear from you! I would also be happy to share my approach/architecture in more detail.

My Setup (Q1 2020)

Mon, 06 Jan 2020 00:00:00 -0600

Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

Home Setup

At home, I switch between three devices. My daily driver is a Razer Blade 15 Advanced (Late-2019). It’s primarily used for productivity tasks, software development, photo editing, virtual labs, gaming and game development (GameMaker Studio 2). The full specification can be found below:

Razer Blade 15 Advanced (Late-2019)
2.6GHz 6-Core Intel i7-9750H (Turbo up to 4.5GHz)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
1TB Samsung 970 EVO Plus PCIe NVMe SSD
Intel UHD Graphics 630
NVIDIA GeForce RTX 2080 Max-Q 8GB GDDR6
15.6-inch 4K OLED Touch DCI-P3 HDR400 Display (3840x2160)

16-inch MacBook Pro with Retina Display (NOV-2019)
2.4GHz 8-Core Intel i9-9980HK (Turbo up to 5.0GHz)
32GB 2666MHz DDR4 RAM
1TB PCIe SSD
Intel UHD Graphics 630
AMD Radeon Pro 5500M 8GB GDDR6
16-inch IPS P3 Display (3072x1920, 500nits)

Alongside my notebooks, I have a custom-built PC. The specification (listed below) is a couple of years old but still provides excellent overall performance. Although released in 2018, the GeForce RTX 2080 Ti is still a phenomenal graphics card, enabling high-performance Ray Tracing gaming.

Asus Maximus VIII Hero Alpha (Intel Z170)
Intel Core i7-6700K 4.0GHz @ 4.6GHz
Noctua NH-D15 CPU Cooler
32GB Corsair Vengeance LPX DDR4 PC4-24000C15 (15-17-17-35)
1TB Samsung PM981 M.2 PCI-e NVMe SSD
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case
Bowers & Wilkins MM-1 Speakers

In total I have three displays, two 27-inch Dell U2718Q IPS HDR 4K monitors used for productivity and colour accurate work, as well as one 27-inch Dell 144Hz G-Sync monitor used for gaming. The exact model numbers are highlighted below:

2x Dell U2718Q IPS HDR 4K (3840x2160 / 60Hz / 5ms / HDR10)
1x Dell S2716DG TN G-Sync (2560x1440 / 144Hz / 1ms)

My workspace is kept fairly minimal, with all monitors mounted on third-party stands to help maximise desk space. The Razer Blade 15 Advanced and MacBook Pro connects to the monitors via a StarTech Dual-DisplayPort to ThunderBolt 3 Adapter, which provides a single-cable solution.

As you can see, for peripherals, I primarily use the Logitech Craft, Logitech MX Anywhere 2S and Logitech Brio Ultra HD Pro. The keyboard and mouse were selected as they include “Easy-Switch”, which makes it simple to toggle between multiple Bluetooth devices.

The image below provides a closer look at the two Dell U2718Q IPS HDR 4K monitors and peripherals.

The far end of the desk provides access to the Dell S2716DG G-Sync monitor, which can be pushed flush against the wall. You can also see the Bowers & Wilkins MM-1 Speakers, which are connected to the PC and Mac. The flat desk edge provides a clean surface for additional peripherals, such as my flight controller (Saitek X-55) and steering wheel (Logitech G29).

As highlighted in the image below, the PC is hidden under the desk, with the case door exposed providing easy access to the components.

The desk and matching pedestals are from Three Counties Office Furniture. They are designed for corporate use, therefore are hard-wearing and include integrated cable management. The dual-monitor arm is from CMD.

Finally, I use a Herman Miller Setu chair, which includes the Kinematic Spine to control resistance and automatically support your weight as you recline.

Work Setup

At work, I use the Razer Blade and/or MacBook Pro (depending on the workload). To provide consistency with home, I leverage the same monitors (Dell U2718Q IPS HDR 4K), display adapter, and peripherals (e.g. Mouse, Keyboard, and Webcam).

Architecture Community

Wed, 11 Dec 2019 00:00:00 -0600

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

The previous articles have focused on design, technology, positioning and delivery. This article will highlight my proposed enterprise architecture methodology, including the organisation structure and processes. The article builds upon concepts I explored in my 2016 series, focused on Enterprise IT Architecture.

These articles are not required reading, simply shared for reference.

Introduction

As per the 2011 Wall Street Journal article, “Software is Eating the World”. To some extent, every business, across every industry, must now consider itself a technology business. Therefore, every business strategy must inherently prioritise a digital business model.

As referenced by Simon Sinek, as part of the popular The Golden Circle framework, Information Technology has the potential to have a profound impact on the “HOW” and “WHAT”, powering the business vision (WHY).

Referencing my previously defined business characteristics, I believe that any business with an emerging digital business model must look to leverage the expertise and experience that undoubtedly exists within IT. In parallel, IT organisations must look to maximise their value contribution by proactively prioritising business outcomes. Therefore, a generic vision for IT could read:

*“Partner to create value through innovative digital products, services and insights, that help to ."*

To enable this vision, a direct and continuous connection must be established between the digital products/services/insights and the business outcome. I would position a “product mindset” as a key philosophical approach, evolving from the traditional “project mindset” that is common across IT.

Project Mindset: A way of funding and organising the delivery of a specific output with a fixed end date.
Product Mindset: Designed to continually create value throughout the lifecycle of a given product.

This philosophy would incorporate key industry concepts, such as product ownership, product teams, iterative planning, continuous improvement and continuous investment.

In short, instead of focusing on a fixed output, a product mindset would focus on the outcome, which would have a profound impact on how the business organises, governs and funds IT.

The diagram below highlights my proposed IT organisation design, supporting 170+ IT users.

Enterprise prioritisation and a focus on maximising value would help ensure that resources were effectively utilised, directly and holistically supporting key business outcomes. Outlined below is a summary of each team, including the high-level roles and responsibilities.

Business Relationship Management (BRM): Accountable for product ownership (including product vision, backlog, lifecycle, communications), as well as enterprise governance, prioritisation and budgeting. BRM would represent the “digital vision” as part of a given business outcome, fostering the required business relationships, expertise and context. This holistic enterprise approach supports my belief that IT must aim to maximise its value contribution, being positioned as a differentiator and driving competitive advantage.
SolutionOps: Accountable for application solution design, delivery and operations, across all business functions, supported by an enterprise Application Managed Services (AMS) provider. SolutionOps would incorporate purchased (Commercial of the Shelf) and custom-developed applications, following a “DevSecOps” culture, with a focus on Continuous Delivery, Testing and Quality. SolutionOps would also be accountable for maintaining Line of Business Apps (commodity/highly-industrialised), allowing for the consolidation of common practices (e.g. release management, regression testing, data integrations and data obfuscation). In theory, this approach would drive economies of scale and efficiency, reducing the risk of duplication, as well as promoting standardisation and re-use through techniques such as innersourcing.
DataOps: Accountable for enabling an automated, process-oriented methodology that facilitates the design and delivery of Enterprise Data and Analytics. DataOps would include data architecture, engineering and stewardship, ensuring data is discoverable, consumable and reusable, as well as enabling advanced data concepts such as Predictive Analytics, Machine Learning and Artificial Intelligence. This dedicated focus on data respects the importance of data as a critical business asset.
TechOps: Accountable for the design, delivery and operations of enterprise IT foundations, including infrastructure, platforms and local site operations. TechOps services would be made consumable via the Automation (e.g. Cloud Direct and Cloud Brokered), enabling autonomy for SolutionOps and DataOps. This approach would help to maximise investments, as well as embed consistent information security, quality and privacy controls.
IT BusinessOps: Accountable for the enterprise management of financials, suppliers, resources, organisational change and the portfolio. IT BusinessOps would work alongside representatives from each team (BRM, SolutionOps, DataOps, TechOps, Information Security), looking to measure and maximise value across processes, policies, investments, financial planning, etc.
Information Security: Accountable for establishing and governing information security requirements, as well as protecting enterprise confidentiality, integrity and availability. Similar to Privacy and Quality, Information Security would be positioned as an embedded practise, therefore, this dedicated team would focus on providing specific expertise and external council, supporting the implementation of the previously described Zero Trust IT security model.

Although the organisation design highlights a horizontally split (enterprise teams), the actual work would occur vertically (product teams).

Product teams would be formed organically, based on the enterprise backlog, prioritisation and budget.

Each product team would include appropriate business expertise (split across Customer, Product, Operations and Corporate), as well as a cross-section of skills, covering product ownership (BRM), application (SolutionOps) and data (DataOps) engineering, as well as consultancy from domain experts (TechOps and Information Security).

The ability to scale elastically would be achieved through the Application Managed Services (AMS) provider, providing access to engineer and specialist resources.

Architecture Community

Building on my 2016 article “Enterprise IT Architecture”, the Enterprise Architecture Community (emphasis on the community) would be a virtual team, including representation from across the organisation.

The community of architects, each with specific expertise (e.g. business, solution, data and domain architects), would ensure enterprise architecture remains relevant, practical and delivering against a clear value proposition. Each architect would represent their area of expertise, providing thought leadership and contributing towards the enterprise outcome. This approach positions architecture as a core part of the end-to-end product lifecycle (not a disconnected team and/or set of processes), which helps to ensure strong business continuity and situational awareness.

The goal of the Enterprise Architecture Community (EAC) would be to provide a “north star” for the enterprise, through the creation of principles, declarations and clear technology positioning via an enterprise reference architecture. Each member of the community would be accountable for distributing the enterprise direction, recognising that architecture as an embedded process, not restricted to those with “architect” in their title. This approach would promote the strategic direction of IT and the business by proactively triggering the right architecture conversations.

The diagram below highlights my proposed Enterprise Architecture methodology, with key components being owned by Enterprise Architecture Community, leveraged by a product team (specifically engineers) during the design and delivery of a new solution.

As highlighted in the article “Modern IT Ecosystem”, I have defined four IT principles for each architecture domain (e.g. Business, Information, Application and Technology). These IT principles have been designed to be easy to consume, remember and adopt, targeting all of IT. The article also includes IT declarations, which build upon the IT principles, providing additional context.

To help ensure a direct and continuous connection between IT and the business outcomes, business capability mapping would be positioned to highlight all required capabilities across the value chain (e.g. Research & Development, Manufacturing, Commercial and Global Services). Any enterprise capability (leveraged by multiple parts of the value chain) would be mapped to the Enterprise Reference Architecture and positioned against key technology investments. This process would help to maximise the return on investment, whilst enabling standardisation and promoting re-use.

Recognising the number of enterprise capabilities, I have split them into the following categories to simplify consumption and maintenance.

Security / Risk
Customer Management
Employee Management
Supplier Management
Process Management
Document Management
Data Management
Asset Management
Engineering
Communications
Enterprise Infrastructure

Finally, I have defined the following process to govern enterprise architecture, placing an emphasis on the previously described IT principles, IT decelerations and reference architecture.

The Enterprise Architecture Community would be positioned as an enabler, not a barrier (something that adds unnecessary bureaucracy). Therefore, the process promotes and incentivises architecture alignment, rewarding the product teams with autonomy (no review required), resulting in speed to value and agility.

Conclusion

I believe this lightweight/embedded approach to enterprise architecture promotes the right behaviours, avoiding the trap of creating an “Ivory Tower”, where the value proposition of architecture would be quickly lost.

The approach does not follow a specific architecture methodology (e.g. TOGAF, etc.) Instead, I have taken the best bits from multiple sources, rightsizing based on my previously defined business characteristics.

VMware Workstation Pro

Sun, 08 Dec 2019 00:00:00 -0600

Since purchasing a Razer Blade 15 Advanced, I have been using Windows 10 Pro as my primary operating system. Although I am no stranger to Windows, it has been over sixteen years since I last used it as my daily driver.

As highlighted in the article “macOS and Windows”, the transition to Windows was fairly seamless. However, I have had a few challenges with my development workflow, which on macOS utilised a combination of Docker and VMware Fusion. I would do the majority of my development locally, leveraging Docker natively installed on macOS, but also have a range of virtual machines (e.g. Fedora, Ubuntu, Windows 10) running to support specific activities.

In theory, the same setup is viable on Windows 10. However, unfortunately VMware Workstation Pro (the Windows alternative to VMware Fusion) can only launch a Virtual Machine if Hyper-V is disabled. This would be fine, except the default install of Docker for Windows requires Hyper-V.

This frustrating “catch-22” means that you cannot run Docker and VMware Workstation Pro simultaneously. If you attempt to do so, you will receive the error “VMware Player and Device/Credential Guard are not compatible”.

Outlined below are three options to mitigate this issue.

Option One: Hyper-V Only

Hyper-V is the native Microsoft Hypervisor built into Windows 10 Pro. Knowing that Docker is built on Linux Container technology, it leverages Hyper-V to run containers on Windows. This makes sense, as Hyper-V is performant, efficient and natively available, allowing Docker to focus on their code (instead of the Hypervisor). There are third-party projects that attempt to make Docker run on a VMware Hypervisor, however, as these are community developed, I am not convinced they can be trusted for business workloads.

Knowing that Hyper-V is a hypervisor, the first option would be to use Hyper-V to host all virtual machines? This would remove the need for VMware Workstation Pro and therefore avoid the conflict, mitigating the issue.

Hyper-V is an excellent hypervisor, especially when acting as a backing service and/or managing server-side workloads. Unfortunately, the client experience has some frustrating limitations.

The initial setup and networking configuration for guest operating systems can be cumbersome.
Access to virtual machines will need to occur via RDP (or equivalent), which will impact some native features of the guest operating system (e.g. audio, video, graphics, etc.)
Installation of native drivers on a virtual machine is complex and in some cases, not possible.
Limitations associated with dynamically attaching and removing external devices from the virtual machine.
Specific limitations related to running Linux virtual machines, impacting core services (e.g. sound, etc.)

In short, Hyper-V on Windows 10 Pro is essentially a “free” embedded hypervisor, primarily used to facilitate specific Windows 10 tasks. VMware Workstation Pro is a paid, standalone product, focused exclusively on hosting virtual machines. Therefore, it should be no surprise that VMware offers the superior experience.

With this in mind, although I could probably make Hyper-V work, I do not feel replacing VMware Workstation Pro with Hyper-V is the right option for me.

Option Two: VMware Only

The second option would be to leave Hyper-V disabled, enabling VMware Workstation Pro and installing Docker on a guest operating system running in a virtual machine.

This option assumes that Hyper-V is not required for any other service, as well as places a higher emphasis on the virtual machine. For example, to maximise productivity, it would likely make sense to shift your entire development workflow to the virtual machine. This could bring some advantages (the use of native Linux), but would also impact performance and require you to perpetually maintain a second operating system.

I prefer this option over option one, helped by the fact that I usually maintain a Fedora virtual machine for other purposes. However, there is something about working for long periods within a virtual machine that feels inefficient and cumbersome, likely caused by the additional abstraction layer between the user input and underlying hardware (audio, video, etc.)

Option Three: Hyper-V / VMware Switching

The final option would be to simply enable or disable Hyper-V depending on the situation.

In my scenario, I would need to decide to run Docker or VMware Workstation Pro, but not both simultaneously. Although this would be frustrating, I rarely execute commands at the same time, therefore making the option viable.

The most efficient way to disable Hyper-V is to run the following command as an administrator.

bcdedit /set hypervisorlaunchtype off

Once complete (image below), the system will need to be restarted.

To reenable Hyper-V, simply run the following command as an administrator.

bcdedit /set hypervisorlaunchtype auto

Again, once the command has completed (image below), the system must be restarted

On a performant system, the overall process should only take a few minutes, but can still be inconvenient, especially if you have multiple applications open.

Conclusion

Today there is no perfect option. I prefer options two or three, however, they both have drawbacks.

In the future, I hope that Windows Subsystem for Linux (WSL) will permanently remediate this issue, allowing Docker to leverage WSL 2 as the backend. In theory, this would remove the need for Hyper-V, replaced by a native Linux kernel.

It would appear this capability is in development (which is great news), but currently only available as a “tech preview”, when combined with a Windows 10 Pro insider build (19018 or higher). I tend to avoid natively installing pre-release software on my primary system, therefore will wait for the general release to complete my testing (expected in April 2020).

Razer Blade

Mon, 18 Nov 2019 00:00:00 -0600

In February 2019, I temporarily switched from a MacBook Pro to a Lenovo ThinkPad X1 Extreme as my daily driver.

Although partly driven by work, I was also growing increasingly frustrated by Apple and the direction they were taking the Mac and macOS. As stated in my previous article, I liked the ThinkPad X1 Extreme, especially the performance and keyboard. Unfortunately, the poor trackpad and subsequent release of the 16-inch MacBook Pro enticed me back to the Mac.

Although the result of this experiment could be considered a failure, there was a lot to like about the Windows ecosystem, therefore I continued to hunt for a laptop that would better meet my expectations (which I appreciate are higher than most).

This led me to Razer Blade 15 Advanced.

It is fair to say that the Razer Blade is heavily inspired by the MacBook Pro, with a full unibody aluminium chassis, minimalist styling and large trackpad. However, arguably Razer has improved upon the design, delivering a 4K OLED touch display, a wider selection of ports (USB-A, Thunderbolt 3, HDMI, Mini-Display Port), upgradable components (Memory, SSD and Wireless), as well as NVIDIA graphics options (GeForce GTX 1660 Ti to Quadro RTX 5000).

These improvements would all count for nothing if the build quality was poor. Thankfully, this is not the case, as the Razer Blade delivers comparable build quality to Apple, with a ridged, dense construction and refined edges.

Building on Razer’s heritage as a gaming company, the Razer Blade is still a “gaming” laptop at its core, but thankfully far better balanced than the competition, with a focus on subtle design, size, weight and battery life. I am not aware of any other gaming laptop on the market that I would consider viable for my workload.

Specification

The Razer Blade series includes multiple models, with customisable options primarily focused on graphics, storage and display. I selected the RTX 2080 Max-Q, 512GB SSD and 4K OLED model. This premium selection commands a high retail price (£3149.99), however, following a few days of research, I was able to purchase one for £2249.00 (still expensive, but more in-line with my expectations).

Although the factory-built Razer Blade specification is good, I took the opportunity to upgrade the memory and storage, selecting 64GB Corsair Vengeance Series 2666MHz RAM (£289.99) and a 1TB Samsung 970 EVO Plus PCIe NVMe SSD (£191.73).

The installation of memory and storage was very simple, through the removal of ten Torx T5 screws, the bottom panel provides clear access to the components.

The full specification of my Razer Blade can be found below.

Razer Blade 15 Advanced (Late-2019)
2.6GHz 6-Core Intel i7-9750H (Turbo up to 4.5GHz)
64GB Corsair Vengeance Series 2666MHz DDR4 RAM
1TB Samsung 970 EVO Plus PCIe NVMe SSD
Intel UHD Graphics 630
NVIDIA GeForce RTX 2080 Max-Q 8GB GDDR6
15.6-inch 4K OLED Touch DCI-P3 HDR400 Display (3840x2160)
3x USB 3.2 Gen 2 (USB-A), 1x Thunderbolt 3 (USB-C)
1x HDMI, 1x Mini- Display Port
Intel Wireless-AX200 (802.11 a/b/g/n/ac/ax)
Bluetooth 5

The full price of the laptop (after the upgrades) was £2730.72, which is still cheaper than the retail price from Razer for this model. It is also favourable when compared against an equivalent Lenovo ThinkPad X1 Extreme (£2700.00) and 16-inch MacBook Pro (£3600.00).

Build Quality

As previously stated, the Razer Blade is very well built, certainly the best I have seen in the Windows ecosystem.

Thanks to the use of OLED technology, the display is something special! It is bright, with the iconic OLED deep blacks and “colour pop”. It is also factory calibrated and colour accurate, but the accuracy can be impacted by changes in peak brightness (which is disappointing).

The keyboard is good (similar to the new 16-inch MacBook Pro) and the trackpad uses Windows precision drivers. Knowing that the trackpad was my main point of contention with the ThinkPad X1 Extreme, I am pleased to report that (in my opinion) the Razer Blade is much better. Not only is the trackpad bigger (similar to a 13-inch MacBook Pro), but also more responsive (although palm rejection is not perfect). Unfortunately, it still falls a long way short of an Apple trackpad, which remains the gold standard.

Performance

Unsurprisingly, the Razer Blade performs very well, offering more than enough power to support any workload (including modern games with Ray Tracing). I have included a few benchmarks below:

PCMark 10: 4279
3DMark Time Spy: 7627
ATTO Disk Benchmark: Average 2700MB Read / 2500MB Write

It should be noted that these scores are real-world numbers, with no overclocking and all of my default software loaded. The laptop is also BitLocker enabled, therefore the PCMark and ATTO scores include a 10% penalty due to the software-based full drive encryption.

Alongside the raw performance, battery life is respectable. When under heavy load (e.g. virtual machines, video editing, gaming), the battery life is obviously impacted (only achieving a few hours). However, I am pleased to report that when focused on productivity tasks and coding, I was able to use the device for the duration of an Atlantic flight (around seven hours). It is certainly no match for my 16-inch MacBook Pro, but the other benefits help to balance the equation.

Connected to the battery, it is worth mentioning the charger, which is quite large (230W) and includes a braided lead which can be difficult to keep tidy. The charger also leverages a proprietary connector and the laptop cannot be charged via USB-C, which is both disappointing and restrictive. Therefore, although travelling with the laptop is easy, managing the charger can be cumbersome, making me miss the (aeroplane compatible) 96W 16-inch MacBook Pro USB-C charger.

Conclusion

I have been incredibly impressed with the Razer Blade. I would have never imagined I would purchase a “gaming” laptop as my daily driver.

However, I believe Razer has achieved the impossible, delivering a laptop that includes a subtle, sophisticated design, portable, powerful and upgradable. I cannot state how great it felt to physically upgrade the laptop post-purchase, something that is rarely an option with modern technology (especially within the Apple ecosystem).

In short, I would highly recommend the Razer Blade!

Line of Business Apps

Fri, 08 Nov 2019 00:00:00 -0600

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

Within this article, I will highlight my proposed positioning for key Line of Business Applications, specifically Enterprise Resource Planning (ERP), Human Capital/Resourcing, Customer Relationship Management (CRM), Regulated Content and Data Management and Service Integration and Management (SIAM).

Introduction

The terminology “Line of Business Application” is not consistently defined, in the context of this article, I will use the term to describe a “mission-critical” application that supports a key business capability.

The typical business value chain includes many highly industrialised processes, covering Research & Development, Manufacturing, Commercial and Global Services. For example, Enterprise Resource Planning (ERP), Human Capital Management (HCM), Customer Relationship Management (CRM), etc. These are critical processes to any business, but could also be considered a commodity.

With this in mind, my philosophy would be to prioritise highly scalable, efficient and reliable services, instead of looking to differentiate where there is minimal competitive advantage.

Enterprise Resource Planning (ERP)

With the previously defined business characteristics in mind, I would position SAP S/4HANA as my Enterprise Resource Planning (ERP) solution, used to manage and track core business processes, resources (cash, materials, production capacity) and operations. The diagram below outlines which capabilities would be supported by SAP, highlighting core components (dark blue) and bolt-ons (light blue).

SAP can be an incredibly complex beast, therefore my preference would be to embrace the capabilities found in S/4HANA, avoiding (where possible) custom configuration.

My single instance of SAP would be hosted by Virtustream, who have a wealth of experience deploying and supporting SAP for large enterprise customers within their specialist (purpose-built) cloud.

I did explore the use of a Public Cloud provider (e.g. Azure) for SAP hosting but prioritised Virtustream based on the following value positioning.

Hosting/Basis: Virtustream deliver unified (one-vendor) SAP Hosting and Basis.
Full Stack: Virtustream would be accountable for the full technology stack and therefore not reliant on a third-party Public Cloud (e.g. Azure).
Experience/Expertise: Outside of SAP, Virtustream has the most experience deploying/supporting SAP.
Cost Effective: Virtustream leverage a unique consumption-based commercial model, delivered via fine-grained measurement of compute, memory, network and storage resources.
Hybrid Multi-Cloud: Virtustream is Public Cloud agnostic (reducing lock-in and protecting against data transit costs).
Architecture Alignment: Virtustream partner with Equinix, who I have positioned as my Colocation Data Centre provider.
Performance: Virtustream would peer with the Equinix ECX Fabric, enabling a cost-effective fibre cross-connect.
Dell Partnership: Virtustream are part of Dell Technologies, maximising my previously described investments across Dell Server, Dell Client, VMware and RSA.

With Virtustream, the business would get a fully-managed SAP instance (SaaS-like) and the benefits of a cloud (utility compute) cost model, combined with cost-effective “LAN-like” performance, that would not penalise for data transit.

The diagram below provides a high-level overview of my proposed Virtustream SAP Hosting architecture.

As depicted, the proposed architecture includes multiple user interaction points, including traditional VPN/VDI via the Colocation Data Centre infrastructure, as well as Direct Internet Access (DIA) for Cloud-Native (Web/API) interactions (embracing SAP Fiori).

Human Capital/Resourcing

I would position Workday as my unified platform for Human Capital/Resourcing, covering Human Capital Management (HCM), Human Resource Management (HRM) and Human Resource Planning (HRP).

Human Capital Management (HCM): A set of practices related to people resource management. These practices focus on the business need to provide specific competencies, covering workforce acquisition, workforce management and workforce optimisation.
Human Resource Management (HRM): The strategic approach to the effective management of people in a business, enabling a competitive advantage.
Human Resource Planning (HRP): The process of forecasting the future human resource requirements of the business and optimising the existing human resource capacity to meet specific business requirements.

Workday would support the following key business capabilities:

Recruiting
Onboarding
Payroll
Timekeeping
Benefits
Compensation
Talent Management/Succession
Organisation Management (Joiners, Movers, Leavers)
Performance Management
Reporting (HR, Headcount, Trending Workforce data, etc.)

Workday is a SaaS platform, with a strong focus on integrations (Workday API Documentation). Alongside SAP, it would become a primary source system of my proposed Identify Access Management (IAM) architecture.

Customer Relationship Management (CRM)

I would position Veeva CRM and Veeva Network for Customer Relationship Management (CRM) and Customer Master Data Management.

Veeva CRM is a vertical built on Salesforce.com, delivering specific capabilities for life sciences (the industry I selected as part of my previously defined business characteristics).

Over the years, Veeva CRM has become the “de facto” standard for Customer Relationship Management across life sciences, delivering advanced mobility, multichannel capabilities and real-time insights for sales teams. Looking outside of life sciences, if a clear market leader does not exist, I would explore Salesforce.com Sales Cloud and/or Microsoft Dynamics (depending on the business requirements and wider investments).

When combined with my proposed endpoint architecture, Veeva CRM would provide an easy to use, rich user experience that would aim to improve productivity for field teams that have limited “office/admin” time.

Veeva Network provides a highly integrated Customer Master Data Management capability, enabling localised business applications and a global data model. When enabled alongside Veeva CRM, it would deliver a seamless end-to-end experience, with no third-party integrations to manage.

Veeva Network also includes a wide range of standard integrations, for example, SAP Concur for expense reporting, which would help to streamline key business processes, whilst improving quality and transparency.

Veeva Network would form part of my wider Enterprise Data & Analytics architecture, which I plan to cover in a future article.

Finally, Veeva Approved Email and Veeva Cloud Loop Marketing (CLM) would be positioned to facilitate multi-channel engagement.

Regulated Content and Data Management

Building on my Veeva CRM and Veeva Network investments, I would position Veeva Vault as my enterprise content management platform. Veeva Vault would manage both content and data, enabling end-to-end processes across commercial, medical, clinical, regulatory, quality and safety. As per my previous defined business characteristics) these processes would be critical to help ensure regulatory compliance.

The Veeva Vault Platform includes multiple applications, specifically:

To maximise the return on investment, I would position the use of the following applications, supporting multiple business functions (R&D, Manufacturing, Commercial and Global Services).

Vault Clinical Operations: Supporting the study process and method control, providing a single system of record for shared CTMS, TMF and study start-up content.
Vault Quality Suite: Supporting the Quality/Control processes, covering Quality Plan, Action Tracking, Change, Deviation, Complaint, Audit, Lab Investigation, Product Safety Assessment (PSA) and Notification to Management (NTM). Veeva Quality Suite would also act as the authoritative source for quality documentation/content (e.g. Procedures, Policies, Validation, Reports, etc.)
Vault RIM Suite: Supporting the planning, execution and oversight of all regulatory activities within a single, unified Regulatory Information Management (RIM) application.
Vault Safety: Supporting the collection, management and real-time oversight of adverse events.
Vault PromoMats: Supporting commercial and scientific disclosure, providing digital asset management, review and distribution capabilities.

Similar to Veeva CRM and Veeva Network, Veeva Vault is a SaaS platform (adhering to my previously defined IT principles and IT declarations).

Service Integration and Management (SIAM)

To enable effective Service Integration and Management (SIAM), the following IT capabilities would need to be established and maintained.

Service Integration Governance: Define, establish and continuously adapt the service integration governance.
Service Integration Organisation: Develop and manage a distributed organisation supporting changing business requirements.
Business Requirements: Manage business demand and develop a service portfolio in alignment with business requirements.
Tools and Information: Manage distributed information and the integration tool solutions.
Providers and Contracts: Select an appropriate provider portfolio and to manage the providers according to the outsourcing contracts.
End-to-end Services: Understand and manage the business services end-to-end, covering business and IT services.

Recognising the critical importance of IT Service Management (ITSM) as part of Service Integration and Management (SIAM), I would position ServiceNow as my unified ITSM/SIAM platform. ServiceNow would support the following key capabilities:

Configuration Management
Problem Management
Change Management
Incident Management
Knowledge Management
Request Fulfilment Management
IT Asset Reconciliation (Hardware and Software)
IT Cost Management
Business Process Tracking (IT Business Management)
User Effort/Satisfaction (KA Surveys)

As outlined above, leveraging the integrations provided by Workday, ServiceNow would compliment specific HR processes, providing workflow capabilities for HR/Ethics and Compliance Case Management, as well as host the previously described Information Portal.

Conclusion

As outlined in the article “Modern IT Ecosystem”, my goal was to target SaaS providers for all highly-industrialised/commodity capabilities, regardless of business criticality.

I believe the capabilities and positioning outlined in this article demonstrate how this could be achieved (even within a regulated industry), enabling “best in class” services, with a focus on the business requirements (not the underlying technology).

Recognising the additional benefits of SaaS, specifically reduced implementation time, simplified support and continuous updates, I believe my “Cloud First” strategy would dramatically improve business agility, enabling a better outcome for the customer.

Fedora 31

Sat, 02 Nov 2019 00:00:00 -0500

In October, Fedora 31 was officially released.

As long-time Linux user, I was eager to test the new release of Fedora (my preferred Linux distribution), therefore I took the opportunity to install it as the primary operating system on a Dell Precision 5540. At work, the Dell Precision 5540 is positioned as our “high Performance” laptop, therefore Linux is a common requirement to support specialist workloads.

The full specification of the Dell Precision 5540 can be found below.

Dell Precision 5540
15” 1920x1080 100% sRGB Non-Touch Display
2.6GHz 6-Core Intel Core i7-9850H (Turbo 4.6GHz)
32GB DDR4 RAM
512GB M.2 PCIe NVMe SSD
Intel UHD Graphics 630
NVIDIA Quadro T1000 4GB GDDR5 (ISV Certified)
Intel Dual Band (802.11ac) Wireless + Bluetooth 5.0
TPM 2.0 + Windows Hello

The Dell Precision 5540 is officially certified to run Ubuntu and thankfully the installation of Fedora 31 went smoothly. The only change required was to switch the “SATA Operation” from “RAID to “AHCI” in the BIOS. Without this change, the Fedora installation wizard could not detect the internal SSD.

Once complete, I began the process of installing my common software. As outlined in the article “Flatpak and Snappy”, the majority of the applications I use support cross-distribution packages.

The steps outlined below, highlight the commands I run to install my common software, making Fedora suitable for daily usage (personal and work).

Before any additional software installation, I always ensure the current software catalogue is up to date. This can be completed via GNOME Software or using the following terminal command.

sudo dnf upgrade --refresh

The only Operating System customisation software that I install is “GNOME Tweaks”, which provides access to advance GNOME features and extensions. For example, the ability to modify the default appearance, keyboard and mouse, etc. To install “GNOME Tweaks”, run the following terminal command.

sudo dnf install gnome-tweak-tool

For better or worse, I still use Google Chrome as my primary browser. Due to some technical limitations, Chrome is not available as a cross-distribution package. However, it can be easily installed via the DNF package manager.

sudo dnf install fedora-workstation-repositories
sudo dnf config-manager --set-enabled google-chrome
sudo dnf install google-chrome-stable

When installed from the official Fedora repository, Google Chrome is automaticlly configured as a “Managed Browser”, meaning you will see the message “Your browser is managed by your organisation”. This is because Fedora applies two Chrome policies that aim to mitigate specific SSO limitations. These policies can be reviewed by navigating to “chrome://policy/”. Although not required, I prefer to remove these policies using the following terminal command.

sudo dnf remove fedora-chromium-config

Flatpak is installed by default with Fedora 31, however, it must be enabled before any software installation.

flatpak remote-add --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo

Once enabled, I install the majority of my software via flathub. For example, the command “flatpak install flathub com.slack.Slack” will install “Slack”.

flatpak install flathub org.libreoffice.LibreOffice
flatpak install flathub com.wps.Office
flatpak install flathub io.atom.Atom
flatpak install flathub com.visualstudio.code
flatpak install flathub com.github.bcedu.vgrive
flatpak install flathub com.spotify.Client
flatpak install flathub org.videolan.VLC
flatpak install flathub com.slack.Slack
flatpak install flathub org.gimp.GIMP
flatpak install flathub fr.handbrake.ghb
flatpak install flathub org.openshot.OpenShot
flatpak install flathub org.kde.kdenlive
flatpak install flathub fr.natron.Natron
flatpak install flathub org.eclipse.Committers

Unfortunately, not all of my software is available via Flatpak, therefore I also install and enable Snappy. It is important to note that you must log out after running each command.

sudo dnf install snapd
# Log Out
sudo ln -s /var/lib/snapd/snap /snap
# Log Out

I install the SimpleNote snap, which is not currently available as a Flatpak.

sudo snap install simplenote

To ensure compatibility with Microsoft Office, I install the True Type font package (e.g. Arial, Georgia, Times New Roman, etc.) using the following terminal commands.

sudo su

sudo dnf install curl cabextract xorg-x11-font-utils fontconfig

sudo rpm -i https://downloads.sourceforge.net/project/mscorefonts2/rpms/msttcore-fonts-installer-2.6-1.noarch.rpm

Once complete, Fedora 31 is ready to go!

Overall, everything installed successfully and Fedora 31 is running well on the Dell Precision 5540. I have not had any issues with drivers and all core components (e.g. graphics, wireless) operate as designed, without any additional user intervention. Although this is exactly what you would expect, it is not always the case with Linux.

Stability has been good, without any major system crashes, but I do get occasional glitches. For example, mouse delays are not uncommon and I receive “kernel oops” upon boot (related to the graphics driver). These minor issues have not impacted my daily usage, but keep Fedora from feeling “polished”.

Performance and Resource Management are both acceptable, but not exceptional. This has become a theme with Fedora, which is not the best performing Linux distribution. At idle, Fedora 31 consumes 4% CPU and 1.5GB of Memory, which is fine in a laptop with 6-cores/12-threads and 32GB, but something to consider if running on a lower specification.

Overall, Fedora 31 is a little underwhelming as an update but delivers what I have come to expect from a new release. As a GNOME user, Fedora is still my preferred Linux distribution, something that I don’t anticipate changing any time soon.

Unified Communications

Wed, 02 Oct 2019 00:00:00 -0500

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

Within this article, I will highlight my proposed Unified Communications and Collaboration (UCC) architecture, describing my philosophy, key technology decisions and positioning. In the interest of brevity, this article will reference “Unified Communications and Collaboration” as UCC.

Introduction

In the broadest sense, UCC encompasses all forms of communication and collaboration across a network, ranging from one-to-one engagement to public broadcasts.

Over the years, the techniques and technologies used to enable UCC have evolved dramatically, but the goal has remained largely unchanged.

“Unify human and device communications, streamlining business processes and improving user productivity.”

In 2014, like Jive, Chatter and Yammer before it, Slack was positioned as the technology that would “kill email”, enabling a true UCC experience. Inevitably, this has yet to happen…

In my opinion, looking for “one technology to rule them all” is a futile task, recognising that UCC is very subjective and context-specific. For example.

Subjective: One person may like email, while another may prefer instant messaging, it comes down to personal preference and working style. This is also influenced by the specific generation (Generation Z, etc.)
Context-Specific: Different communication channels are effective in different scenarios. For example, a Webinar (live voice and video) is a more effective channel than email for a virtual event.

With this in mind, my proposed architecture would not target a single technology. Instead, I would create a UCC brand, positioning multiple technologies that complement each other.

I believe this approach would promote diversity within the workplace (encouraging different working styles), whilst still providing a framework for effective collaboration.

Architecture

To support my positioning, I have defined a model that groups the different forms of collaboration into four types.

Personal Collaboration: Collaboration between a limited number (less than 10) of known participants. Collaboration could include Asynchronous Messaging, Synchronous Messaging, Voice, Video, Screen Sharing, Document Sharing and Document Co-authoring.
Team Collaboration: Collaboration within a team or group (less than 50), where all participants are known. Collaboration could include Asynchronous Messaging, Synchronous Messaging, Voice, Video, Screen Sharing, Document Sharing and Document Co-authoring.
Community Collaboration: Collaboration within a community (more than 50), where all participants are known. Collaboration could include Asynchronous Messaging, Live Video Streaming, Town Hall Meetings, Webinars, Digital Signage, Information Portal and Document Sharing.
Open Collaboration: Collaboration with an open community of any size, where the participants are unknown. Collaboration could include Asynchronous Messaging, Synchronous Messaging, Live Video Streaming, Webinars and Call Centres.

The diagram below builds upon the model, mapping my proposed UCC architecture across the four types of collaboration.

As depicted, I would position the following technologies.

Personal Collaboration

Outlook: Email.
Teams: Messaging, Voice, Video and Screen Sharing.
Zoom: Voice, Video and Screen Sharing.
OneDrive: Document Sharing and Document Co-authoring.

Team Collaboration

Teams: Messaging, Voice, Video and Screen Sharing.
Zoom: Voice, Video and Screen Sharing.
OneDrive: Document Sharing and Document Co-authoring.

Community Collaboration

Workplace: Messaging and Live Video Streaming.
Zoom: Town Hall Meetings, Webinars and Digital Signage.
ServiceNow: Information Portal.
OneDrive: Document Sharing and Document Co-authoring.

Open Collaboration

Public Social Networks: Messaging, Live Video Streaming and Webinars.
Genesys Pure Cloud: Call Centres.

UCC can be a contentious area, as the technologies are widely used, but prone to error (specifically real-time voice and video). Therefore, outlined below are some of the key considerations that informed my architecture positioning.

Microsoft Office 365

Building upon the investments highlighted in the article “Hybrid Multi-Cloud”, I have positioned Microsoft Office 365 (specifically Outlook, Teams and OneDrive) to enable a highly integrated experience, which would include the core productivity tools (e.g. Word, PowerPoint, Excel, etc.)

Although the diagram does not specifically reference SharePoint, it is important to recognise that many of the Office 365 services are powered by SharePoint, including Teams and OneDrive. Therefore, although I would not position the use of SharePoint directly, it would exist within the proposed architecture, enabling capabilities such as unified search, etc.

Teams would be positioned as the primary technology for Messaging, Voice, Video and Screen Sharing. However, Microsoft has a “checkered past” when it comes to voice and video, with a history of poor execution (e.g. Office Communicator, Lync, Skype for Business, etc.)

As a result, I would position Zoom as an alternative option for users who require the best possible voice and video experience. In my experience, Zoom offers a “best in class” user experience, prioritising performance, reliability and multi-platform support.

To maximise this investment, Zoom would also be positioned for specialist capabilities, such as Town Hall Meetings (10,000 viewers), Webinars and Digital Signage.

In 2012, Microsoft acquired Yammer, which is a social networking service designed for enterprise communication. Unfortunately, over the past seven-years, Microsoft has failed to capitalise on this acquisition, resulting in a disjointed user experience. With the new focus on Teams, it is unclear how Yammer will be positioned in the future, potentially remaining a standalone service or perhaps being merged into Teams (we might learn more in November at Microsoft Ignite).

Due to this uncertainty, I would not position Yammer as a core service, instead I would position Workplace as a cost-effective alternative. Workplace is built by Facebook and therefore delivers an intuitive/comfortable user experience for social collaboration. Thanks to Facebook’s estimated 2.45 billion active users, Workplace requires very little training, dramatically reducing the organisational change impact.

The screenshot below highlights Workplace in action, providing a familiar experience for anyone that has used Facebook.

Although running on a separate technology-stack, Workplace and Facebook share the same DNA, enabling common capabilities such as News Feed, Groups, Live Video, etc. Also, Workplace enables specific enterprise capabilities, such as Single Sign-On, Org Charts, OneDrive and Zoom Integrations, as well as Safety Check (supporting crisis communications).

Information Portal

A traditional Information Portal (AKA Intranet) may be considered antiquated, but it is easy to overlook the power of a simple page that provides links to important resources (e.g. Help Desk, Payroll, HR, Training, etc.)

I have previously positioned ServiceNow as my IT Service Management platform, therefore would look to maximise this investment by leveraging their “Employee Service Centre” as an Information Portal.

The screenshot below is an example of an Information Portal built using ServiceNow, highlighting key capabilities such as Search, Service Requests, Case Management, Virtual Agents, etc.

The ServiceNow Information Portal would be simple and cost-effective to implement, maintain and support. It would also provide direct access to all requests and knowledge articles stored in ServiceNow, as well as native integration with HR services such as Workday.

OneDrive and Digital Assistant

Positioned across all collaboration types (e.g. Personal, Team and Community) would be two key services, OneDrive and a Digital Assistant.

OneDrive would provide unified content synchronisation and collaboration, tightly integrated with the wider productivity and UCC architecture (e.g. Teams, Workplace, etc.) It would also provide a consistent foundation for document sharing and document co-authoring, as well as central control pane for governance, security and data loss prevention.

The Digital Assistant represents a vision of the future, enabled via an intelligent chatbot, which could be accessed from any of the UCC technologies (e.g. Teams, Workplace, ServiceNow, etc.) The goal of the Digital Assistant would be to streamline business processes, by providing answers to commonly asked questions, as well as automating daily tasks. For example, a user could ask “how many holiday days do I have left” or “submit a holiday request for Friday”.

Telephony

Although telephony is one of the most mature forms of network-based communication, it is often still the most complex and costly to implement at scale.

As previously referenced, I have positioned Teams as my primary technology for Messaging, Voice, Video and Screen Sharing. Building upon this strategy, I would enhance the Microsoft 365 E5 Phone System feature by integrating with a Direct Routing service to enable a PSTN breakout at required sites.

To facilitate the PSTN interconnection I would position AudioCodes as the Session Border Controller (SBC) and Analogue Gateway supplier. The proposed SBC architecture would be hosted in Azure and/or turnkey appliances at the local sites (driven by in-country regulations for the provisioning of Direct Inbound Dial numbers).

Finally, maximising my existing investment, I would position CenturyLink to integrate the telephony components, enabling a single dial plan across the enterprise, delivered to the business “as-a-Service”.

This approach to telephony would compliment my winder UCC architecture, delivering a singular interface for all voice requirements.

Conclusion

In conclusion, I believe my proposed UCC architecture supports my goal to enable personal preference, while still delivering a consistent, simple, user experience. This would be accomplished through a unified brand and carefully positioned technologies that complement each other, providing a cost-effective, reliable and extensible foundation.

BitLocker

Mon, 30 Sep 2019 00:00:00 -0500

On 24th September, Microsoft released a software update for Windows 10 (KB4516071) which modified the default behaviour of BitLocker.

BitLocker is the embedded full drive encryption technology built into Windows, offering protection against data loss if the physical hardware becomes compromised. I personally always enable BitLocker on portable Windows devices, as well as the equivalent full drive encryption technologies on macOS (Apple File Vault) and Linux.

Historically, where available, Windows would enable hardware-based encryption, which offers better performance than the software-based equivalent. For example, when buying a Samsung SSD, Samsung provides the option to enable hardware-based encryption using its own method, designed specifically for the drive itself.

Although more performant, this approach relies on the hardware provider to ensure the encryption method is secure. Unfortunately, security researchers have discovered that not all providers can be trusted, potentially putting the data at risk and voiding the value of full drive encryption.

As a result, Microsoft has made the decision to make software-based encryption the default for BitLocker moving forward (existing drives will not be changed).

Although I fully understand and respect the response from Microsoft (protecting the user should always be the first priority), it does mean that by default, anyone looking to leverage BitLocker will (by default) be sacrificing performance. When combined with the performance penalties associated with other vulnerabilities such as Meltdown and Spectre, this outcome is a little disappointing.

As an example, in the article “ThinkPad X1 Extreme”, I benchmarked (PCMark 10 Windows 10 without drive encryption. Following the new update form Microsoft, I enabled BitLocker and re-ran the benchmark. The results can be found below.

BitLocker Disabled: 4716
BitLocker Enabled (Software-Based): 4259

Although not completely scientific, the benchmark highlights a performance penalty of approximately 10%, which is in line with my expectation of software-based encryption.

If you have enabled BitLocker, but unsure if hardware-based encryption has been enabled, simply run the following command as an administrator from the command line.

manage-bde.exe -status

If hardware-based encryption is enabled, it will be clearly stated as part of the read-out. For example, the image below does not specifically state hardware-encryption.

Finally, assuming you trust your hardware provider, it is possible to overwrite the default BitLocker setting. To achieve this outcome, I recommend referring to your hardware providers documentation, as the process may require third-party software. For example, Samsung drives can be managed using their “Samsung Magician” software.

Flatpak and Snappy

Sun, 22 Sep 2019 00:00:00 -0500

Package managers provide a way of packaging, distributing, installing, maintaining, and removing applications. Packages contain metadata, such as the applications name, description, version, vendor, checksum, and a list of dependencies.

In the Linux ecosystem, package management is very important, as applications are downloaded and installed from repositories, often requiring specific dependencies to run correctly.

Unfortunately, package management on Linux can be confusing, as different Linux distributions prioritise different package managers. For example, Linux distributions based on Debian prioritise the Debian Package Management System (Dpkg), while Linux distributions from Red Hat prioritise the Red Hat Package Manager (RPM).

To add additional complexity, there are multiple ways to interact with a package manager. For example:

Debian Package Management System (Dpkg)

Advanced Packaging Tool (APT): A front-end command line package management system for the Debian package manager.
Aptitude Package Manager: Another popular front-end command line package management system for the Debian package manager.
Synaptic Package Manager: A GUI package management tool for the Advanced Packaging Tool (APT).

Red Hat Package Manager (RPM)

Yellowdog Updater, Modified (YUM): A command-line package manager, providing an interface to the Red Hat Package Manager (RPM).
Dandified Yum (DNF): A fork of the YUM package manager and the new default for Fedora (22 onwards).

I am most familiar with YUM and DNF, but even after years of use, I still find managing packages cumbersome (hence I created a DNF Cheat Sheet.

Thankfully, over the past few years, package management across the Linux ecosystem has been slowly evolving, driven by the desire to deliver cross-distribution packages.

Introducing Flatpak and Snappy, which are both cross-distribution package managers.

Flatpak

Flatpak was developed as part of the freedesktop.org project and was originally called xdg-app. Key features include:

Containerised: Software isolation, improving reliability and security.
Cross-distribution: Can be distributed across multiple Linux distributions.
Stable Platform: Tightly coupled runtimes, providing a consistent foundation.
System Permissions: Granular control over system resources and permissions.
Dependency Control: Bundle all core-dependencies.
Unified Distribution: Distribution via FlatHub.
Open-Source: Flatpak is run as an independent and open-source project.

Flatpak is supported by 21 Linux distributions, including Ubuntu, Fedora, Mint, etc. Packages can be discovered, downloaded and installed from FlatHub.

Snappy

Snappy was developed by Canonical (e.g. the developer of Ubuntu). The packages are called snaps, enabled by the snapd tool. Key features include:

Read-Only: read-only, signed content, improving control and security.
Self Contained: Decoupling applications from the underlying system.
Cross-distribution: Can be distributed across multiple Linux distributions.
Stable Platform: Include a compressed filesystem that is mounted before execution.
System Permissions: Application communications occur via approved interfaces.
Desktop and IoT: Support for desktop and IoT Linux use cases.
Simplified Distribution: Distribution via Snapcraft.
Dependency Control: Bundle all core-dependencies.

Snaps are supported by 41 Linux distributions, including Debian, Fedora, etc. Unsurprisingly, Snappy comes pre-installed on Ubuntu. A popular source to discover, download and install snaps is Snapcraft, as well as the Snap Store app.

My Applications

Earlier this year, I shared my list of commonly used Linux software. Building on that post, the table below highlights which applications are available via Flatpak and/or Snappy.

As you can see, in the context of my workload, I have reasonably good coverage across Flatpak and Snappy. It is also worth mentioning that I found the search on FlatHub to be far more accurate/effective than Snapcraft.

Conclusion

There are other cross-distribution package managers, for example, AppImage. However, the majority of my usage has been with Flatpak and Snappy.

It is a little ironic that multiple technologies have emerged in an attempt to reduce the fragmentation of package management across the Linux ecosystem. Personally, I prefer Flatpak, as I am predominantly a desktop user and believe their “independent” position better serves the Linux community.

With that said, Flatpak and Snappy can be used simultaneously (which is what I do on Fedora), therefore, at least these new cross-distribution package managers do not conflict with each other.

In summary, I am an advocate for the use of cross-distribution package managers, although ideally there would be one independent open-source option, instead of competing technologies. Over time, we might see consolidation or clearer positioning, where specific package managers are positioned for desktop, IoT, etc.) Finally, I am excited to see how features such as “application isolation” and “decoupling” can help enable new paradigms like immutable operating systems (e.g. Fedora Silverblue).

Identity Access Management

Tue, 17 Sep 2019 00:00:00 -0500

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

Within this article, I will highlight my proposed Identity Access Management (IAM) architecture, describing my philosophy, key technology decisions and positioning.

Introduction

As previously described, I have positioned a Zero Trust IT security model. Zero Trust is a holistic approach to IT security, which goes beyond the traditional “trust but verify” and “moat/castle” strategies. Although these traditional strategies are still common within enterprise businesses, they primarily target environments where the business has end-to-end ownership, management and control of all IT services.

In a modern IT ecosystem, it is very common for IT services to be distributed (e.g. SaaS Services, Cloud Hosting. etc.) Therefore, the users, services and data could be anywhere and everywhere, making the task of IT security increasingly complex.

The answer to this challenge is not to perpetuate the inadequate strategies (e.g. moat/castle), but instead look to implement a new strategy, where the network is always assumed to be hostile, meaning internal and external threats exist at all times.

With this in mind, Zero Trust implements new principles, techniques and technologies to help localise and isolate threats. For example, the following principles provide the foundation for a Zero Trust architecture.

The network is always assumed to be hostile, meaning external and internal threats exist on the network at all times.
Network locality (moat/castle) is not sufficient for deciding trust. No user or device is automatically trusted.
Micro-segmentation concepts create secure enclaves, limiting network and application flows between workloads, reducing the “blast radius” of an attack.
Every device, user and network flow is authenticated and authorised, preferably leveraging Multi-Factor Authentication (MFA).
A least-privilege access philosophy (need-to-known basis), where users are only granted the required access to complete their job (nothing more).
Policies must be dynamic and calculated from as many sources of data as possible, enabled via a central management control plane.

Enabling these principles requires new technologies and techniques, such as Modern Identity Access Management, Microcore, Segmentation and Deep Visibility, which aim to provide a structured approach to threat identification, as well as limiting the impact of a breach.

At the foundation, a Zero Trust architecture must be able to complete strict identity verification for every user and device trying to access a business resource, regardless of whether they exist within or outside of the network perimeter.

Therefore, from an IT security perspective, Identity Access Management (IAM) becomes one of the most important and powerful technologies used to protect the business.

Architecture

Identity Access Management (IAM) aims to define and manage the role and access privileges of a user (e.g. employees, contractors, customers) and the circumstances in which they would be granted (or denied) specific privileges.

The primary goal of Identity Access Management (IAM) is to create a single digital identity per user. Once that digital identity has been established, it must be maintained, modified and monitored throughout the user’s access lifecycle.

The diagram below outlines my proposed Identity Access Management (IAM) architecture, including the provisioning flows.

To enable this architecture, I would position the following technologies:

Workday, SAP: Multiple decoupled source systems (specifically HR and ERP) synchronise data with the Master Identity Management System, providing flexibility and agility, specifically covering mergers, acquisitions, divestitures, etc.
RSA Identity Governance Lifecycle (IGL): A unified Master Identity Management System, managing the lifecycle of identities, facilitating the Joiner/Mover/Leaver (JML) processes and providing continuous compliance capabilities.
Global Identity Repository: A connected data source, providing secure/scalable access to user data via a RESTful API.
Active Directory: Directory services, providing authentication and authorisation services for the Colocation Data Centres, Edge Computing and legacy federation.
Azure Active Directory: Directory services, providing authentication and authorisation services for applications in the cloud (SaaS, Public Cloud), as well as modern federation.

Positioning RSA (part of Dell Technologies) and Microsoft complements my Hybrid Multi-Cloud and Device-as-a-Service strategies, whilst also simplifying the Identity Access Management (IAM) architecture through capability consolidation. For example:

RSA IGL would focus on Access Certification, User Provisioning, Policy Automation and Role Management, whilst providing self-service capabilities and enterprise visibility across the access lifecycle.
Microsoft Active Directory and Azure Active Directory enable key authentication patterns, including SAML, OAuth 2.0, OpenID Connect, MAPI, WS-Federation, Kerberos, LDAP, etc.

As highlighted as part of my proposed Automation architecture, I have positioned ServiceNow for IT Service Management (ITSM). Although not depicted, ServiceNow would also facilitate identity processes, specifically request fulfilment and change control (approvals), as well as support any workflow automation requirements.

NOTE: I plan to cover my proposed IT Service Management architecture in a future article, which will include additional details regarding the positioning of ServiceNow.

Authentication Patterns

The diagram below outlines a range of authentication patterns, including their preference based on the proposed architecture.

As you can see from the diagram, the user would have a single digital identity, with applications/services prioritising modern authentication.

With the previously defined business characteristics in mind, traditional authentication patterns cannot be completely ignored (specialist applications in R&D and Manufacturing). In these scenarios, the user would have a non-standard username, which would conform to the least common denominator (e.g. numbers only). In theory, through applications modernisation, these traditional authentication patterns would reduce over time.

Local Authentication: Local (decentralised) databases used for identity would not be supported within the ecosystem.
Kerberos: Kerberos is an authentication protocol that works based on tickets to allow nodes communicating over a non-secure network to prove their identity to one another in a secure manner. Kerberos would not be preferred, but technically viable within the proposed architecture.
LDAP: The Lightweight Directory Access Protocol (LDAP) is an open protocol for accessing and maintaining distributed directory information services over a network. Similar to Kerberos, LDAP would not be preferred, but technically viable within the proposed architecture.
Direct Connect: Specific services that require a direct connection, potentially via an insecure protocol. These services would not be directly accessible to the Internet and include an additional level of authentication before establishing a connection.
Modern Authentication: Modern Authentication leverages tokens, sent to an authentication provider. This removes the need for the username and password to be sent over the network. For example, SAML, OAuth 2.0, OpenID Connect, MAPI, WS-Federation, etc. Preferred for all applications and services.
Endpoint Authentication: Aligned with “Modern Authentication”, enabling password-less authentication via a gesture (e.g. facial recognition, iris scan, fingerprint, etc.) Preferred for all endpoints (clients/mobiles).

My holistic vision would be to enable a password-less ecosystem, where modern authentication patterns would be utilised by all applications, services and endpoints.

Thanks to my positioning of Microsoft Windows 10 and Apple iOS (Clients/Mobiles), password-less could be enabled by default, leveraging technologies such as Windows Hello and Apple Face ID/Touch ID.

Windows Hello is compatible with any service that supports Fast Identity Online (FIDO). Therefore, as Windows Hello matures, the use of password-less would increase over time.

Conclusion

In conclusion, I feel strongly that any modern IT ecosystem must target a Zero Trust IT security model, where Identity Access Management (IAM) would play a critical role.

This approach would help to ensure the ecosystem is protected, without limiting the ability to innovate and leverage cutting-edge technologies. The option to enable password-less authentication for the majority of use cases would be an exciting prospect, recognising that IT can rarely improve security and the user experience at the same time.

Device-as-a-Service

Wed, 11 Sep 2019 00:00:00 -0500

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

Within this article, I will highlight my proposed Device-as-a-Service (endpoint) architecture, describing my philosophy, key technology decisions and positioning.

Introduction

As described in the article “Modern IT Ecosystem”, I have positioned a Zero Trust security model, which aims to localise and isolate threats through microcore, segmentation and deep visibility.

This model also extends to the endpoints (e.g. desktops, laptops, tablets, smartphones), which I would treat as a commodity (untrusted). To enable this outcome, the enterprise security controls would be layered throughout the end-to-end ecosystem, with solutions being secure by design (secured at the source), whilst complemented by a robust Identity Access Management architecture and a clear “least privilege access” strategy.

As a result, my proposed endpoint architecture would prioritise productivity, delivering a “consumer-like” experience.

For example, enterprise endpoints are commonly overloaded with third-party security software, generally resulting in multiple running agents. These agents continuously consume system resources (processor and memory), as well as often require kernel-level access, which (ironically) opens a high-risk attack vector.

As an alternative, I would leverage the native capabilities of the specific device (no third-party software), looking to embrace the embedded security mechanisms and controls. Not only would this dramatically simplify the architecture, but it would also improve the end-user experience, making it comparable to a consumer purchase from Dell, Apple, etc.

This approach also supports my goal of making the ecosystem device-agnostic, where any endpoint (e.g. Windows, macOS, Linux, ChromeOS, iOS Android) could be utilised with minimum effort. I feel this would be an important architecture tenant, recognising the rapid growth of connected devices (e.g. Sensors) and new forms of Human-Computer Interaction (e.g. Mixed Reality Headsets).

Considering this security model and device-agnostic strategy, you might assume that I would position “Bring Your Own Device (BYOD)”. Over the past decade, BYOD has become a popular trend, however, the true value is often reduced/lost due to the inherent complexities regarding security, privacy and local law. As a result, many respected analysts have rebranded BYOD as “Bring Your Own Disaster”.

Therefore, I would position a “Choose Your Own Device (CYOD)” strategy, where the enterprise retains ownership of the device itself (clear separation between personal and business), but still provides flexibility for users to personalise their experience.

Device-as-a-Service

As previously stated, I see the endpoint as a commodity. As a result, I would treat the processes (e.g. provisioning, lifecycle, break-fix) supporting the endpoint “as-a-Service”. The “as-a-Service” paradigm is well established with software (e.g. SaaS, Public Cloud), but is more complex when physical hardware is involved.

Thankfully, the major personal computer providers (e.g. Dell, Lenovo) all offer Device-as-a-Service and/or PC-as-a-Service capabilities, where they aim to combine hardware, software, lifecycle, break-fix and financing as one all-encompassing service.

In this scenario, the business simply provides the required specification (e.g. hardware, software) and configuration (e.g. settings, policies), with all other logistics being managed by the personal computer provider. The high-level diagram below outlines the process.

A key enabler of this process would be the ability to configure the builds at the point of use. Historically, enterprise businesses would have to create and maintain a build (image), which would include a specific version of the operating system, as well as all base drivers, configuration and software. The provisioning and maintenance of this build (or builds) could be frustratingly complex, as well as resource-intensive and time-consuming.

With Windows 10, Microsoft introduced a new capability known as Windows Autopilot, which removes the need to manage a traditional build, allowing devices to be automatically configured upon delivery.

Windows Autopilot

Windows Autopilot is a capability designed to pre-configure, provision, repurpose and recover devices. I would position the use of Windows AutoPiot, looking to simplify my end-to-end device lifecycle management process, supported by a certified personal computer provider.

Windows Autopilot uses an OEM-optimised version of Windows 10, which comes pre-installed on the device. This removes the need for the enterprise to create and maintain a custom build (image).

At the point of deployment, instead of re-imaging the device, the OEM-optimised version of Windows 10 is transformed into a “business-ready” state. This includes the required configuration (e.g. settings, policies) and software (e.g. Office 365, Chrome).

Post-deployment, Windows Autopilot could be used to re-purpose the device by leveraging “Windows Autopilot Reset” or to support break-fix events.

As the name suggests, Windows Autopilot is a Windows-only feature (targeting Windows 10). However, recognising the framing of this series, I have assumed that 98% of the endpoints would be running Windows 10, with the remaining devices using a combination of macOS (1%) and a variant of Linux (1%).

Microsoft Intune, SCCM and JAMF

Microsoft Intune is a unified endpoint management solution, covering Windows, macOS, iOS and Android. I would position Intune as my primary endpoint management solution, used to automate provisioning, policy management, application delivery and updates.

Unfortunately, Intune is not yet feature-complete, especially when targeting legacy operating systems, as well as macOS and Linux. To ensure end-to-end compatibility, Microsoft System Center Configuration Manager (SCCM) and JAMF Pro would be positioned to compliment Intune.

Across the ecosystem, the use of SCCM and JAMF should be very limited (less than 5%). As Intune matures, the reliance on SCCM and JAMF should reduce further.

Hardware Architecture

As previously stated, I would target a “Choose Your Own Device (CYOD)” strategy, positioning Dell as my Autopilot certified personal computer provider for Device-as-a-Service (maximising the investments made as part of the Colocation Data Centre architecture).

To simplify the deployment of hardware, I would leverage a set of user personas (e.g. Office, Sales Representative, etc.) Each persona would be assigned default hardware; however, users would have the option to update their hardware selection to better support their business requirements.

Based on the current Dell range, I would position the following hardware, which would obviously evolve overtime:

Business Laptop

Dell Latitude 5400
14” 1920x1080 Non-Touch Display
1.6GHz 4-Core Intel Core i5-8365U (Turbo 4.1GHz)
16GB DDR4 RAM
256GB M.2 PCIe NVMe SSD
Intel UHD Graphics 620
Intel Dual Band (802.11ac) Wireless + Bluetooth 5.0
TPM 2.0 + Windows Hello

2-in-1 Laptop

Dell Latitude 7400
12.3”/14” 1920x1280/1920x1080 Touch Display
1.6GHz 4-Core Intel Core i5-8365U (Turbo 4.1GHz)
16GB DDR4 RAM
512GB/512GB M.2 PCIe NVMe SSD
Intel UHD Graphics 620
Intel Dual Band (802.11ac) Wireless + Bluetooth 5.0
TPM 2.0 + Windows Hello

High-Performance Laptop

Dell Precision 5540
15” 1920x1080 100% sRGB Non-Touch Display
2.6GHz 6-Core Intel Core i7-9850H (Turbo 4.6GHz)
32GB DDR4 RAM
512GB M.2 PCIe NVMe SSD
Intel UHD Graphics 630
NVIDIA Quadro T1000 4GB GDDR5 (ISV Certified)
Intel Dual Band (802.11ac) Wireless + Bluetooth 5.0
TPM 2.0 + Windows Hello

Apple Mac

Apple MacBook Pro
13” 2560×1600 P3 Colour Space Non-Touch Display
2.4GHz 4-Core Intel Core i5-8279U (Turbo 4.1GHz)
16GB DDR4 RAM
256GB NVMe SSD
Intel Iris Plus Graphics 655
802.11ac Wireless + Bluetooth 5.0
Apple T2 + Touch ID

All devices come with a minimum 1920x1080 resolution display, 16GB RAM, NVMe SSD, as well as hardware security and biometric authentication. This would provide a very strong foundation, delivering performance, sustainability and security across the entire range. The positioned High-Performance laptop also includes ISV certified graphics (NVIDIA Quadro T1000), providing support for specialist workloads.

Alongside the laptop range, I would also position two desktop computers (Dell Optiplex 3070 micro and Precision Workstation 3431). The use of desktop computers would be limited, but likely still a requirement for certain user personas. The Precision Workstation 3431 comes equipped with a Xeon Processor, ECC RAM and NVIDIA Quadro Graphics (ISV Certified), making it a very versatile computer, that could support the most demanding workloads.

Finally, regarding mobility, I would position the Apple iPhone and Apple iPad Air. At this time, I would not position Google Android as a standard mobile offering, however, this might change as the market matures.

Software Architecture

As a result of the previously defined security approach (leveraging the native capabilities of the specific device), the initial client software configuration would be incredibly simple.

Windows Desktop/Laptop

Microsoft Windows 10 Enterprise 1903 (AKA 19H1)
Microsoft Office 365 (Word, PowerPoint, Excel, Outlook, Teams)
Google Chrome
Palo Alto Networks Prisma Access

Apple Mac

Apple macOS 10.15.1 (Catalina)
Microsoft Office 365 (Word, PowerPoint, Excel, Outlook, Teams)
Google Chrome
Palo Alto Networks Prisma Access

With Microsoft releasing two major operating systems updates per year, I would aim to remain one version (six months) behind the current build (e.g. 1903 vs. 1909). This would help ensure the business remains current but also protected against unforeseen bugs via a short stabilisation period. The same philosophy would be applied to macOS, waiting for the release of 10.x.1, etc.

As per the previous defined IT principles and IT declarations, the majority of applications and services should be cloud-hosted (SaaS, Public Cloud), accessible over the Internet, via a browser or an API. I would position Google Chrome as the default browser, complimented by Microsoft Edge (Chromium Edition). Any traditional application that requires installation would be downloaded via the web or through Intune.

Only specialised applications running in a Colocation Data Centre would require direct network connectivity via the LAN/WLAN or client VPN. This is where Palo Alto Networks Prisma Access would be utilised to provide full-tunnel encryption.

NOTE: I plan to cover my proposed security architecture in a future article, which will include additional details regarding the use of Palo Alto Networks.

Conclusion

In conclusion, it is exciting to be able to propose an endpoint architecture that would prioritise productivity and user experience, whilst maintaining high standards regarding Information Security, Privacy and Quality. This is a rare combination, which demonstrates the value proposition of a modern IT ecosystem, helping to empower users, optimise operations and improve business agility.

Linux Software

Tue, 02 Jul 2019 00:00:00 -0500

In a recent article, I shared my philosophy to software selection, which supports my requirement to frequently switch between different operating systems.

Recognising that my daily driver is a Mac, I shared a list of my most used macOS applications and the Windows alternatives.

Thanks to the growth of web technologies and frameworks such as Electron, many applications are now cross-platform, providing a relatively seamless experience when working across multiple operating systems.

Knowing that I am also a frequent user of Linux (specifically Fedora), I felt it was worth expanding upon the previous list to include the equivalent Linux applications.

Although the Linux eco-system is still behind macOS and Windows, it continues to mature at a steady pace, including renewed support from major software providers such as Microsoft, as well as many active open-source projects.

As I reflect on the application list, I am comfortable with the coverage provided by Linux. The open-source Electron framework, which is used for the development of desktop GUI applications using web technologies, has helped drive parity across operating systems. For example, Atom, Visual Studio Code, Slack, Teams are all Electron-based applications.

Productivity software (Office) remains an area of frustration. Popular Linux options include OpenOffice and LibreOffice, but if you need 100% Microsoft Office compatibility, you are forced to use Microsoft Office Online. In my experience, the most compatible Microsoft Office option on Linux is WPS Office, but is still far from perfect.

Video editing and visual effects can also be challenging on Linux. Although OpenShot, Kdenlive and NATRON are all very capable applications. I miss the simplicity, performance, and ecosystem of applications like Final Cut Pro and the Adobe Creative Suite. It should be noted that Natron is reliant on the open-source community and has struggled in recent years, making new releases and patches very inconsistent. An application such as Blender is better supported but has a steeper learning curve (especially for simple 2D effects).

The only gap in my workflow is 2D Game Development, driven by the fact that I use GameMaker Studio 2. Unfortunately, it does not look like we will ever receive a native version for Linux, which is a shame, but understandable based on the market share.

Overall, I am confident I could use Linux as my daily driver, but due to my reliance on Microsoft Office, I suspect it would impact my overall productivity. Therefore, Linux remains an operating system that I use for specific tasks, such as software development, containerisation, etc.

Automation

Sat, 08 Jun 2019 00:00:00 -0500

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

In the article “Hybrid Multi-Cloud”, I shared my application/data hosting strategy, covering SaaS, Public Cloud, Colocation Data Centres and Edge Computing. This strategy emphasises modern application architecture but offers enough flexibility to support almost any workload.

To unlock the value of this strategy, automation becomes a necessity, helping to streamline development and operations, improve the user experience, as well as guarantee security, quality and privacy.

This article will highlight my proposed automation architecture, describing my philosophy, key technology decisions and positioning.

Introduction

Historically, when dealing with legacy technologies, the effort to enable automation would often outweigh the value. Thankfully, due to advancements in Cloud, API-Centric Architecture and Software-Defined Infrastructure, automation has not only become simpler but also a prerequisite when looking to unlock the value of a modern IT ecosystem. This is highlighted by the value proposition outlined below.

Improved Speed: Reduce the reliance on manual intervention, streamlining the IT processes from ideation to production.
Improved Agility: Ability to deliver continuous improvements, enhancements and issue resolution.
Improved Autonomy: Self-service for IT provisioning, supporting multiple personas (e.g. Product Team, Engineering Team, etc.)
Improved Flexibility: Ability to complete real-time, ad-hoc releases across the entire IT ecosystem (e.g. Cloud, Colocation DC, Edge Computing, etc.)
Enforced Best Practices: Enforce consistent solution design and developer best practices (e.g. Style Guides, etc.)
Improved Information Security: Centralised governance, including role-based access and monitoring. Proactively mandated and enforce security controls (e.g. Linter, Source Code Analysis, Package Analysis, etc.)
Improved Quality: Proactive service discovery and CMDB Configuration Item creation. Proactively mandated and enforce quality controls, including testing (e.g. Application Scaffolding, Application Manifest, Unity Testing, UI Testing, etc.)
Improved Privacy: Mandated and enforce privacy controls, including data encryption, obfuscation, residency, high-availability and backup.
Cost Optimisation: Removal of duplicate technologies (technical debt) to support IT provisioning, governance budgeting, discovery and change control. Removal of time-consuming manual tasks and associated resources.

To reinforce my position regarding automation, I have embedded key automation concepts and techniques as part of my IT principles and IT decelerations. For example, the three statements below provide a clear direction regarding the desired approach to architecture.

Server Deployments: By default, new IaaS deployments must target immutable infrastructure, including ephemeral (short-lived) and stateless (do not persist a session) servers.
Software-Defined: By default, all cloud services (SaaS, Public Cloud), Colocation Data Centre and Edge Computing must be fully automated, leveraging Infrastructure-as-Code and Software-Defined techniques.
Automated: By default, new application deployments must be automated, leveraging build tools and/or automation services.

A full description of my IT principles and IT decelerations can be found in the article Modern IT Ecosystem.

Automation Requirements

The word “automation” covers a very wide scope, therefore, with the previously defined business characteristics in mind, I have created the following requirements, starting with three personas:

Product Team: IT Users and/or Business Partners. Assume basic IT skills.
Engineering Team: IT Users and/or Service Integrators. Assume moderate/advance IT and software development skills.
IT Admins/Operations: IT Users and/or Service Integrators. Assume moderate/advance IT and software development skills.

These personas would interact with the following automated capabilities, connecting back to the previously stated value proposition.

Provisioning: The ability to create, modify and provision low-level (e.g. Server) and high-level (e.g. DNS) infrastructure, as well as software services.
Governance: The ability to create, modify and maintain groups that control the user’s ability to provision, modify and remove infrastructure and services.
Budgeting: The ability to proactively understand provisioning costs, as well as track the spend over time (based on defined groups).
Discovery: The ability to identify, track and manage all provisioned infrastructure and services, including the creation and maintenance of CMDB Configuration Items.
Change Control: The ability to track and manage change across all provisioned infrastructure and services.
Operations: The ability to trigger housekeeping and maintenance activities, based on pre-defined criteria/events and/or predictive insights.
Testing: The ability to proactively trigger pre-defined tests, covering Unit, Module, Integration, System and UI.
Standards Checks: The ability to proactively trigger pre-defined security, quality and privacy checks.
Documentation: The ability to generate and maintain quality (installation/configuration/change) documentation.

To achieve this outcome, new concepts, techniques and technologies would need to be implemented, including a paradigm shift to Immutable Infrastructure.

Immutable Infrastructure

In a traditional “mutable” infrastructure, services can be modified post-deployment, usually via a manual process. For example, a developer and/or administrator would log in to a server via RDP/SSH and complete a set of manual steps to modify configuration, packages, etc.

Immutable infrastructure is a new infrastructure paradigm, where services cannot be modified post-deployment. If a modification is required, a new service is built from a common image, which inherently includes the required changes. This is only possible, if a modern application architecture is followed, for example, The Twelve-Factor App methodology.

The “Pets vs. Cattle” analogy is often used to describe this paradigm shift, where historically services would be treated like pets (one of kind, long life, managed by hand). Whereas modern (immutable) services would be treated like cattle (one of many, shorter life, managed at scale).

To support the previously defined value proposition, I would prioritise immutable infrastructure, which inherently promotes the use of automation, enabled via technologies such as Infrastructure-as-Code (IaC).

Infrastructure-as-Code

Infrastructure-as-Code (IaC) is about writing code that describes the infrastructure, similar to writing a program in a high-level language. Once defined as code, infrastructure can be managed as if it were software, where resources can be created, removed, replaced, resized and migrated, leveraging automation.

Building upon my Hybrid Multi-Cloud strategy, I would position Terraform and Ansible as part of my proposed automation architecture.

Terraform is a cloud-agnostic Infrastructure-as-Code capability owned by HashiCorp, written in Go using HashiCorp Config Language (HCL). It includes an open-source command-line tool, which can be used to provision (low-level and high-level) infrastructure to multiple hosting environments, including Azure, GCP, AWS, as well as VMWare.

Ansible is an open-source Configuration Management, Deployment and Orchestration tool owned by Red Hat. Ansible leverages Playbooks (YAML files), to execute specific tasks, often focused on server configuration. Similar to Terraform, Ansible is agent-less, supporting a wide range of platforms and hosting environments.

In my proposed architecture, Terraform would be positioned to provision infrastructure, delegating configuration tasks to Ansible. The short video below from VMware provides a good overview as to how/why Terraform and Ansible complement each other.

These two technologies not only adhere to my IT principles (open-source, decoupled, API-first) but also help to enable my Hybrid Multi-Cloud strategy, meeting the requirement to deploy to multiple hosting environments (Public Cloud, Colocation DC and Edge Computing).

Finally, it should be noted, as with any technology Terraform and Ansible have their own set of challenges (e.g. Timely Updates, State Management, Secret Management, etc.) With this in mind, the use of cloud provider-specific Infrastructure-as-Code (e.g. Microsoft ARM Templates would be positioned to mitigate any gaps.

Automation Architecture

Alongside Terraform and Ansible, I would position the following technologies to support the wider automation requirements.

Embotics: SaaS-based Cloud Management Platform (CMP), which supports provisioning (via Terraform), governance (group management and change control), budgeting (proactive cost analysis and chargeback), as well as discovery (CI creation).
GitHub: SaaS-based source code management, providing distributed version control for Terraform, Ansible Playbooks and application source code.
Azure DevOps: Maximising the return on investment with Microsoft, Azure DevOps provides a suite of SaaS-based services covering CI/CD capabilities, as well as test tooling and package management.
ServiceNow: SaaS-based IT Service Management (ITSM), providing change control and configuration management database (CMDB) capabilities.
Monitoring: A suite of tools (e.g. Azure, VMware, Application Performance Management), which monitor the end-to-end infrastructure, applications and data, with the ability to predict and proactively trigger maintenance activities.

The decision to position a Cloud Management Platform (CMP) was driven by the desire to proactively manage governance and budgeting, providing enterprise-level visibility and control. The CMP would also enable autonomy for non-technical personas (e.g. Product Teams) by provisioning infrastructure and services to any hosting environment via the simplified service catalogue. However, behind the scenes, Terraform and Ansible would be doing the heavy lifting, ensuring that the CMP remains loosely coupled, enabling both “Cloud Direct” and “Cloud Brokered” provisioning approaches.

The diagram below highlights how these technologies (alongside Terraform and Ansible) complement each other to enable my proposed automation architecture.

As highlighted in the diagram, the Product Teams primary engagement would be via “user-centric” services, such as Embotics and ServiceNow. Embotics provides a service catalogue for standard services (e.g. Servers, Databases, Web Apps), covering all hosting environments (Public Cloud, Colocation Data Centres and Edge Computing). Where required, ServiceNow would manage all change control, including approvals.

The engineering team would primarily work within GitHub, following the GitFlow branching model. As code is pushed to GitHub, automated tasks would be executed via Azure DevOps following a standard CI/CD practice. These tasks could cover a wide range of capabilities, including linters (e.g. ESLint), testing (e.g. Mocha), security analysis (e.g. Checkmarx), package analysis (e.g. Snyk) and document generation (e.g. Pandoc).

Finally, IT admins and operations would support the hosting environments, leveraging native services (e.g. Azure Portal, VMware vSphere) and tooling (e.g. Azure PowerShell, CLI).

Conclusion

In conclusion, my automation strategy aims to facilitate a wider paradigm shift, where infrastructure would be defined and maintained as software. I believe this approach would unlock tremendous business value, specifically speed, agility, flexibility and cost optimisation, as well as promoting innovation.

Swift Playgrounds

Sun, 02 Jun 2019 00:00:00 -0500

Over the years, I have explored a lot of different programming languages, specifically JavaScript, Ruby, PHP, Python, Apex, GML, etc.

In 2014, Apple released the Swift programming language. I have always been intrigued by Swift but lacked the reason and/or time to start the learning process.

Conveniently, my son (aged 4) is now at an age where he can start to understand the basics of coding. Therefore, I thought I would use this as an excuse for us to explore the Swift programming language together, looking to validate Apple’s claim that Swift is easy to learn.

What is Swift?

Swift is a general-purpose, multi-paradigm, compiled programming language.

As the name suggests, Swift has a focus on speed, specifically the enablement of rapid application development, as well as code performance. For example, when compared against Objective-C, Swift is a clean, expressive and concise language with a simple syntax that makes it easy to read and write. It is also up to 40% faster than Objective-C, with clear community guidance on how to deliver highly-optimised code.

Alongside speed, Swift targets safety, with a strong typing system and robust error handling. This further enables rapid application development, as errors can be identified in real-time, resulting in a shorter feedback loop, reducing the risk of production issues.

In 2015, Apple made Swift open-source, which alongside the continued growth of the Apple ecosystem (e.g. macOS, iOS, WatchOS, tvOS) has contributed to its steady increase in popularity.

Therefore, alongside JavaScript and Python, I believe Swift is a viable programming language for anyone looking to learn to code.

What is Swift Playgrounds?

Swift Playgrounds is a REPL (Read–Eval–Print–Loop) development environment for Mac and iPad. This development paradigm aims to enable a simple, interactive programming environment that takes single user inputs, evaluates them, and returns the result to the user in real-time.

Therefore, Swift Playgrounds can be a powerful education and testing tool, which Apple has incentivised through a collection of highly interactive lessons (grouped as playgrounds).

Swift Playgrounds is available for free and includes a wide range of lessons, which can be completed offline.

Using Swift Playgrounds

Recognising the interactive nature of Swift Playgrounds, it felt like the perfect environment for my son (aged 4) and I to learn Swift together.

Alongside Swift Playgrounds, Apple has produced some excellent supporting materials, covering all ages and levels of experience. To get started, I would recommend reviewing the content found on “Apple Education - Teaching Code” website. Specifically, the “Curriculum Guide”, which structures the leaning, providing useful tips and tricks.

The curriculum states that the “Learn to Code” playground (created by Apple) targets ages 10+, which I feel is a fair guide. However, even at 4, my son and I were able to complete the “Learn to Code 1 (Fundamentals of Swift)” playground, which includes the following lessons.

Chapter 1 - Commands: Students learn about the importance of clear, precise commands.
Chapter 2 - Functions: Students explore the power of functions by grouping commands into a definition they can use over and over.
Chapter 3 - For Loops: Students recognise patterns in the world around them and in their code, and learn how to write more efficient code using loops.
Chapter 5 - Conditional Code: Students explore how Boolean logic helps us make decisions in our everyday lives and in code.
Chapter 6 - Logical Operators: Students explore scenarios that require them to consider multiple factors before they make a decision.
Chapter 7 - While Loops: Students explore the power of a while loop for repeating a section of code multiple times until a condition is met.
Chapter 8 - Algorithms: Students learn how an algorithm is a set of rules and instructions used to solve a problem.

Like most children, my son will look for any opportunity to play with a smartphone or tablet. Therefore, it did not take a lot of convincing to get him involved.

In my opinion, the best part about Swift Playgrounds is the fun and highly interactive visuals, which are structured as a set of increasingly difficult puzzles (like a game). Each puzzle adds a new element or concept, which progressively increase in complexity. For example, you start by issuing one command, before adding multiple commands.

My son immediately understood the concept and (with some gentle guidance) was able to quickly move through the opening chapter, clearly building confidence as he completed each puzzle.

At 4, it is worth noting that my son has only just started school, therefore cannot read or write. When writing code, you would assume this would be a hindrance, however, Swift Playgrounds provides handy keyboard shortcuts (essentially making the puzzles multiple choice).

Therefore, although my son cannot read the words, he can identify key letters (helped by the use of camelCase). For example, with “moveForward()”, he would look for the “m” and “F” vs. turnLeft() where he would identify the “t” and “L”. I would provide support where required and verbalise the words and key concepts as we progressed.

The video below is a short clip of my son working through one of the puzzles.

Overall, we were able to progress through the first five chapters with relative ease, at which point the complexity did increase to a level that required a greater amount of support. Unsurprisingly, as the puzzles became more complex, my son’s interest and enthusiasm began to wane. He was perfectly capable of completing the challenge but lacked the patience to work through the problem. Therefore, I feel Apple’s 10+ age guidance is a good benchmark.

The photo below is my son proudly showcasing his first “function” (excuse the crazy look).

Having reviewed the “Learn to Code 2 (Beyond the Basics)” playground, I do not believe this is viable for my son (yet), therefore we will continue to refine our skills by repeating the fundamentals.

Conclusion

Overall, I commend Apple for their commitment to code and programming, especially their goal of making it fun, engaging and accessible to all ages.

In my experience, the “Learn to Code” playgrounds successfully support the development of core coding skills and programming vocabulary (helped by the concise and expressive syntax of Swift). For example, my son cannot code, but he does now understand some of the basic concepts (even if it is unconsciously). These concepts are critical when looking to learn more advanced programming techniques, as well as support the growth of other core skills such as maths, reading and writing.

The only area of improvement I have observed would be the jump from core coding concepts to real-world application development. For example, the first time a new programmer attempts to apply core concepts to develop a real application, using a full IDE like Xcode, can be very daunting.

Apple provides resources to support application development, but they switch from fun and interactive learning to heavyweight documents and/or README files associated with code repositories. In my opinion, this can be a big jump, which I would like to see Apple attempt to mitigate (maybe via a simplified Xcode for iOS).

Finally, it worth noting that Swift Playgrounds also supports third-party playgrounds, which is great to see! I have downloaded the Sphero and Tello playgrounds, which support direct interaction with physical hardware, allowing anyone to programme a drone, etc. My only criticism is that these playgrounds are not always well maintained (not the fault of Apple), meaning that certain functionality can stop working with newer released of iOS, etc.

In conclusion, I am very impressed with Swift and Swift Playgrounds and would certainly recommend the programming language to anyone looking to learn to code.

My Setup (2019)

Wed, 29 May 2019 00:00:00 -0500

Update - Since writing this article I have upgraded to the 16-inch MacBook Pro (NOV-2019). As a result, this article has been updated to reflect the changes.

Final Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

Home Setup

At home, I switch between three devices. My daily driver is a 16-inch MacBook Pro with Retina Display (NOV-2019). It’s primarily used for productivity tasks, software development, and photo editing. The full specification can be found below:

16-inch MacBook Pro with Retina Display (NOV-2019)
2.4GHz 8-Core Intel i9-9980HK (Turbo up to 5.0GHz)
32GB 2666MHz DDR4 RAM
1TB PCIe SSD
Intel UHD Graphics 630
AMD Radeon Pro 5500M 8GB GDDR6
16-inch IPS P3 Display (3072x1920, 500nits)

I also have a 15-inch Lenovo ThinkPad X1 Extreme, which is a secondary device and primarily used for work, allowing me to natively interact with the Windows eco-system. I have previously shared my experience using the ThinkPad X1 Extreme, as well as techniques I use to co-exist across macOS and Windows. The full specification can be found below:

Lenovo ThinkPad X1 Extreme
2.6GHz 6-Core Intel Core i7-8850H (Turbo up to 4.30GHz)
32GB DDR4 2666MHz RAM
1TB M.2 NVMe SSD
Intel UHD Graphics 630
NVIDIA GeForce GTX 1050 Ti Max-Q 4GB GDDR5
15.6” 4K UHD Touch IPS Display (3840x2160, 400nits)

Alongside my notebooks, I have a custom-built PC. The specification (listed below) is a couple of years old but still provides excellent overall performance. It is used for gaming (including VR), game development (GameMaker Studio 2), video editing, after effects and virtual labs.

Asus Maximus VIII Hero Alpha (Intel Z170)
Intel Core i7-6700K 4.0GHz @ 4.6GHz
Noctua NH-D15 CPU Cooler
32GB Corsair Vengeance LPX DDR4 PC4-24000C15 (15-17-17-35)
512GB Samsung 950 Pro M.2 PCI-e NVMe SSD
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case
Bowers & Wilkins MM-1 Speakers

2x Dell U2718Q IPS HDR 4K (3840x2160 / 60Hz / 5ms / HDR10)
1x Dell S2716DG TN G-Sync (2560x1440 / 144Hz / 1ms)

My workspace is kept fairly minimal, with all monitors mounted on third-party stands to help maximise desk space. The MacBook Pro and Lenovo ThinkPad X1 Extreme connect to the monitors via a SoNNeT Technologies Dual-DisplayPort to ThunderBolt 3 Adapter, which provides a single-cable solution.

The image below provides a closer look at the two Dell U2718Q IPS HDR 4K monitors and peripherals.

As highlighted in the image below, the PC is hidden under the desk, with the case door exposed providing easy access to the components.

The desk itself and matching pedestal is from Three Counties Office Furniture. They are both designed for corporate use, therefore are hard-wearing and include integrated cable management.

Finally, I use a Herman Miller Setu chair, which includes the Kinematic Spine to control resistance and automatically support your weight as you recline.

Work Setup

At work, I use the MacBook Pro and Lenovo ThinkPad X1 Extreme (depending on the workload). To provide consistency with home, I leverage the same monitors (Dell U2718Q IPS HDR 4K), display adapter, and peripherals (e.g. Mouse, Keyboard, and Webcam).

BMW 335d M-Sport

Tue, 07 May 2019 00:00:00 -0500

Alongside tech, health, and fitness, I am passionate about motorsport.

I was recently in the market for a new car, replacing a Mercedes C-Class that had served me well for the past three years. The C-Class was purchased after the arrival of my first child, which replaced a less practical BMW 1-Series M-Sport Coupe (E82). There was nothing inherently wrong with the C-Class, but it lacked poise and was not very exciting to drive.

I now have two children under the age of five, therefore my goal was to find a car that would be fun to drive, reasonably economical, as well as practical for daily use with a young family (must fit two car seats and one pushchair). My previous two cars were purchased from new, but I am always open to other options, including second-hand (anything to reduce the depreciation impact).

My research led me to the BMW 335d xDrive M-Sport. Specifically, the 2016 F30 model, which can be found for around £22,000 (RRP £42,000). Recognising the age of my target vehicle, I would be purchasing second-hand, which can be a time-consuming and frustrating process.

To reduce the burden, I engaged Prestige Diesels and Sports Limited, based in Portsmouth. I discussed my desired specification with the team, who used their network (and industry knowledge) to locate a viable option. Once found, they completed a thorough inspection (internal, external and mechanical), before purchasing the vehicle. At which point, I was contacted for a test drive (I had not made any commitment up to this point).

Needless to say, the team had done a great job and the following week I purchased a 2016 BMW 335d xDrive M-Sport Plus in Black Sapphire. It should be noted that Prestige Diesels and Sports only target specific vehicles, usually with “fully-loaded” and/or a rare specification.

I won’t claim to be a photographer, but I have included a few photos below.

The 335d is a four-door saloon, with a reasonable size boot/trunk (480 litres), which is comparable to the C-Class it replaced.

Taken from the “Shadow Edition”, the 335d includes 19-inch Bicolour Orbit Grey 704 M alloy wheels, as well as Bridgestone Potenza S001 run-flat tyres. It should be noted, run-flat tyres can be expensive, costing up to £240 per tyre (including fitting).

The specification of the BMW 335d xDrive M-Sport can be found below.

Layout: Front Engine, All Wheel Drive
Transmission: Eight Speed-Sport Automatic
Engine: N57D30T1 6-cylinder Turbo

The BMW N57 is a family of aluminium, twin-turbocharged straight-6 common rail diesel engines. The engines utilise variable geometry turbochargers and Bosch piezo-electric injectors, producing the following performance results.

Power: 308BHP
Torque: 630 Nm
0 - 60MPH: 4.6 Seconds
Top Speed: 155MPH (Electronically Limited)
Economy: 42MPG (BMW Reference: 52.3MPG)

It was the car’s performance characteristics that intrigued me the most, specifically the compelling performance vs. fuel economy. For example, the table below compares the 335d against the 340i (F30) and M3 (F80).

In short, I believe the 335d hits the sweet spot, offering close to M3 levels of performance, whilst being far more practical (fuel economy, running costs, etc.) Recognising that the 335d is xDrive (all-wheel drive), it is also comparatively easier to drive is varied weather conditions, allowing a less experienced driver to maximise the performance.

The standard “M-Sport” model includes a reasonable base specification. The “Plus” package aims to improve the value by adding Xenon Headlights, Harman/Kardon Speaker System, M-Sport Braking System, etc. My car also includes the “Premium”, “Technology”, and “Media (Professional)” packages. Noteworthy equipment includes:

Active Cruise Control
Auto Stop/Start
Brake Energy Regeneration
Heads-up Display
Speed Limit Display
Lane Assist
Collision Warning
Rain Sensor
Automatic Headlight Activation
Park Distance Control (Front/Rear)
Reversing Assist Camera (Rear)
Professional Multimedia Navigation System
Harman/Kardon Speaker System
Enhanced Bluetooth with Wireless Charging
Wireless Apple CarPlay
Electric Seats and Driver Memory (Front)
Lumbar Support (Front)
Seat Heating (Front)
Split-Folding Seat (Rear)
19-inch Bicolour Orbit Grey 704 M Allow Wheels

This is the first car I have owned to include wireless Apple CarPlay.

The ability to connect and charge wirelessly is a nice benefit. For short journeys, my iPhone never leaves my pocket, connecting automatically a few seconds after starting the car. Once connected, CarPlay is available from the BMW Multimedia system, including access to applications such as Contacts, Messages, Waze, Spotify, YouTube Music, etc. Conveniently, CarPlay applications can also be viewed and controlled from the Heads-up Display. In my opinion, this is the future of in-car entertainment.

Overall the 335d is everything I had hoped it would be! However, as with all things in life, there are a few considerations…

The BMW 3-Series has always been known for its legendary handling. However, the F30 includes electronic (not hydraulic) power steering, which in “comfort” mode can feel a little soft and artificial. Thankfully, “sport” mode is much better, delivering a harder, more connected feel that is closer to what I would expect from a BMW.

Regarding looks, the 335d (even the M-Sport model) is very understated. If de-badged, it would be difficult to identify it against any other F30 3-Series. In general, I like the default style, but feel the duel exhaust design is a little lacklustre. In a perfect world, I would have liked for BMW to take inspiration from the M3 (F80), delivering a duel exhaust design that highlights the cars performance. Another very minor criticism is the height of the car, which (due to the xDrive) rides a little higher than a standard M-Sport 3-Series.

Finally, the Bridgestone Potenza S001 tyres have received mixed reviews online. Although I have not had any major issues with the grip, I have experienced pressure drops, noise and tramlining (which is possibly compounded by the electronic power steering). Replacing all four tyres is a fairly expensive process, but likely a requirement to ensure comparable tread depths (within 2mm) between the front and rear.

In conclusion, I have been incredibly impressed with the BMW 335d xDrive M-Sport Plus. It offers lively performance, whilst still being practical as a family car. Targeting better performance would likely enter “supercar” territory and/or compromise fuel economy, space, etc.

WSL

Fri, 12 Apr 2019 00:00:00 -0500

Web development on Windows can be a frustrating experience, mainly because Windows is not POSIX compliant. As a result, web developers often favour macOS or Linux, which are POSIX compatible (even if Linux is not fully certified).

Since February, I have increased my usage of Windows, which has forced me to explore alternative approaches to support my common workloads (which includes web development). Historically, I have used a hypervisor and/or Docker to provide a viable development environment on Windows. However, each of these options can be frustrating, recognising that they are both reliant on virtualisation.

Therefore, I decided to explore Windows Subsystem for Linux (WSL), which is a compatibility layer for running Linux binary executables (in ELF format) natively on Windows 10.

WSL provides a Linux-compatible kernel interface (containing no Linux kernel code), which supports a GNU userspace, such as Ubuntu, openSUSE, Debian, etc. The userspace provides access to a Bash shell, native GNU/Linux command-line tools and programming language interpreters (e.g. Ruby, Python. etc.) Microsoft does not promote the use of graphical applications, however, this can be partially accomplished using an X11 server running on Windows 10 (e.g. Cygwin X, Xming, etc.)

Due to the use of a compatibility layer (instead of a real Linux kernel), WSL has several limitations. For example, it is not capable of running all Linux software, specifically 32-bit binaries. It is also not possible to install kernel modules and device drivers.

The WSL architecture can be summarised into two parts:

LXSS Manager Service: The LXSS Manager Service interacts with the Windows Subsystem for Linux (via the drivers lxss.sys and lxcore.sys), allowing Windows to launch Linux processes, as well as handle Linux system calls and binary locks during execution.
wsl.exe: The wsl.exe command is used to manage distributions on the command-line. The command can also be used to run Linux binaries from the Windows Command Prompt or Windows PowerShell.

In my experience, WSL works reasonably well. Being able to access the Bash shell from within Windows feels a little strange, but is a step in the right direction for web developers. Unfortunately, due to the use of the “compatibly layer”, WSL is not perfect, limited system call compatibility and the relatively poor IO performance can quickly become a point of frustrating when working with complex web applications.

Recognising WSL is still fairly immature, I hope Microsoft continues to iterate on the architecture, potentially looking to include a full Linux kernel in a future release. This outcome would likely be seen as blasphemy by some Windows evangelists but would overcome the current limitations and potentially provide the best of both worlds (Windows + Linux). If successful, I could see Windows becoming a more popular/common option for web developers, providing a viable (and cheaper) alternative to macOS.

Getting Started with WSL

The steps outlined below, highlight how to install Windows Subsystem for Linux (WSL) using Ubuntu 18.04, without the use of a Microsoft Account or the Microsoft Store.

Open “Control Panel” and search for “Developers”. Enable “Developer Mode” by checking the box highlighted in the image below:

Open Windows PowerShell, ensuring that you select “Run as Administrator”.

Run the following command, which enables WSL.

Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Windows-Subsystem-Linux

Download the approved Ubuntu installation file from Microsoft, using the following URL:

https://aka.ms/wsl-ubuntu-1804

Once downloaded, install Ubuntu.

Open the newly installed Ubuntu application.

Complete the installation by creating a username and password.

WSL is now fully installed and available for use. For more information regarding WSL, I recommend referring to the WSL Setup Guide.

Hybrid Multi-Cloud

Sat, 06 Apr 2019 00:00:00 -0500

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

This article will focus on my proposed cloud architecture, highlighting my philosophy, key technology decisions and positioning.

Public Cloud

Over the past ten years, the Public Cloud market has seen massive growth. I wrote a short article in 2013, highlighting the dominance of Amazon Web Services (AWS). At the time, AWS was a clear leader, however, the market has matured, resulting in three key players: Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP).

As to which Public Cloud provider is “best” depends on multiple factors, specifically, business requirements, user-base, budget, existing investments and supporting architecture. To simplify the landscape, I often use the LEGO Group as a way of describing the differences between the providers.

LEGO = AWS: Like LEGO, AWS is the most popular, with the largest eco-system, highest-rate of innovation and customer engagement. AWS is incredibly versatile, with thousands of services that can be connected to create/support a wide range of solutions. Similar to LEGO, AWS has a strong focus on the eco-system (via its API-Centric strategy), therefore, services tend to target technology-centric customers with software engineering expertise.
DUPLO = Azure: Building on their history as an enterprise-focused company, Microsoft often complement their services with enterprise-specific features. Like DUPLO, the “building blocks” tend to be “larger” and therefore, easier to consume for business-centric customers with limited software engineering expertise. Finally, similar to the connection between DUPLO and LEGO, Azure services are often tightly integrated into the wider Microsoft eco-system (e.g. Office 365).
LEGO Technic = GCP: Google is very much about engineering excellence, with a strong emphasis on custom/proprietary compute, storage and networking architecture (e.g. TPU). Similar to LEGO Technic, Google is often the first to pioneer new/innovative services (e.g. Kubernetes), allowing for the creation of more “exotic” solutions. In regards to market share, Google is in third place, resulting in a more tightly defined eco-system, which attracts customers with a strong appreciation for engineering excellence.

This analogy is far from perfect and is not designed to articulate the value of each provider. It does, however, recognise the different philosophy, target audience and market size, highlighting that a one-to-one comparison does not tell the full story.

Cloud Strategy

As part of my previously described IT principles and IT declarations, I have positioned a clear direction regarding a “Cloud First” strategy, prioritising Public Cloud for applications and data, with a focus on Cloud Native architecture.

The words “Cloud First”, not “Cloud Only”, are important, as they start to promote the idea of a hybrid architecture, with clear prioritisation. Acknowledging the previously defined business characteristics, I specified an established traditional business model, likely resulting in workloads that would not be commercially or architecturally a good fit for Public Cloud. As a result, as part of my modern IT ecosystem, I have positioned multiple highly-converged Colocation Data Centres, alongside Edge Computing at specific sites (primarily R&D and Manufacturing).

Regarding the Public Cloud itself, I anticipate the need for workloads to run across multiple clouds, unlocking innovation and ensuring healthy cost competition. With this in mind, my proposed cloud strategy would be considered “Hybrid Multi-Cloud”.

To facilitate this outcome, I would position a Cloud agnostic, decoupled, automation layer, supported by techniques and technologies such as Software Defined, Infrastructure-as-Code and Immutable Infrastructure, etc. Due to the importance of this area, I plan to describe my proposed automation architecture in a future article.

The philosophy behind “Multi-Cloud” is certainly compelling, however, as with any commercial investment, the value of standardisation should not be underestimated. As a result, I would position Microsoft Azure as my preferred Public Cloud provider. The following two factors have driven this decision:

Although this series assumes no legacy from a technology standpoint, all businesses have a human legacy, with many users (and partners) having a long history consuming Microsoft productivity and collaboration services. Therefore, positioning Microsoft Office 365 would be a logical decision to ensure business continuity and to help reduce the organisation change impact. The use of Office 365 opens the door to a wider commercial deal with Microsoft, allowing the business to maximise the investment across productivity, collaborations and cloud.
With the previously defined business characteristics in mind, I highlighted an established traditional business model and growing digital business model. Therefore, I would aim to capitalise on a business-centric skill-set. For example, businesses operating a traditional business model commonly lack Solution Architect and Software Developer expertise, becoming dependant on service integrators for solution delivery. As outlined above, I believe Azure is best positioned for business-centric customers with emerging software engineering expertise.

With this positioning in mind, the diagram below highlights my proposed application/data hosting strategy, covering SaaS, Public Cloud, Colocation Data Centres and Edge Computing.

Highly-industrialised/commodity workloads would be positioned for SaaS. Examples would include Productivity, Collaboration, Service Management, Customer Relationship Management (CRM), etc.

Public Cloud would be prioritised for all application/data hosting, emphasising “up the stack” services. For example, FaaS and PaaS would be prioritised over IaaS.

I would position Equinix as my Colocation Data Centre host, providing support for workloads that would not be commercially or architecturally a good fit for Public Cloud. For example, specialised capabilities from R&D and Manufacturing, which commonly require proprietary software/hardware. As highlighted in the article “Modern IT Ecosystem, the proposed Colocation Data Centre architecture would be highly-converged, where I would position Dell EMC PowerEdge MX modular servers running VMware Software-Defined Data Centre (SDDC) technologies. This combination would help to maximise the investment (both part of Dell Technologies), as well as provide flexibility to scale compute, storage and network independently.

Edge Computing would support latency-sensitive applications, predominantly at R&D and Manufacturing sites. Where possible, these sites would complement the proposed Colocation Data Centre architecture, leveraging Dell servers (potentially hyper-converged) and VMware SDDC technologies.

Conclusion

In conclusion, I believe my proposed Hybrid Multi-Cloud strategy would provide a strong foundation for an enterprise business, emphasising modern architecture, whilst being pragmatic, providing options for almost any workload.

This balance would provide ultimate business flexibility, whilst continuing to promote modern application architecture, which could be further incentivised via a streamlined governance and custom charge-back model.

SD-WAN

Fri, 01 Mar 2019 00:00:00 -0600

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

Alongside many other critical decisions (e.g. ERP, HRP), I have prioritised the infrastructure foundations, covering the Network, Hosting and Identity Access Management (IAM). This article will aim to provide a high-level overview of my proposed Wide Area Network (WAN) architecture, which I believe presents an interesting opportunity to innovate.

Over the past decade, Multi-Protocol Label Switching (MPLS) has been the standard for enterprise businesses, providing scalable, protocol-independent, any-to-any connectivity. The “magic” behind MPLS is that packets are assigned labels, which are used to make packet-forwarding decisions, without needing to interrogate the packet itself. This approach is very versatile, allowing businesses to implement end-to-end connectivity, across multiple protocols.

MPLS is a service that must be purchased from a carrier (e.g. AT&T, CenturyLink, Verizon, etc.) Depending on the specific network requirements, MPLS services could be very expensive, especially when compared against commodity Internet connections.

With the previously defined business characteristics in mind, I have positioned 50+ physical sites across the globe, including 10+ critical sites that support highly sensitive business processes (e.g. R&D, Manufacturing, etc.) My vision of a modern IT ecosystem places a heavy emphasis on cloud services, covering SaaS (e.g. Workday) and Public Cloud (e.g. Microsoft Azure, etc.)

Although MPLS remains a viable option for the WAN, a new technology known as Software-Defined WAN (SD-WAN) has been gaining market momentum.

SD-WAN is not directly comparable to MPLS, as it is an overlay technology. However, it has the potential to indirectly impact MPLS adoption, due to its ability to aggregate several WAN connections into one software-defined network (SDN).

For example, SD-WAN enables businesses to select inexpensive commodity Internet connections, whilst still maintaining (and even improving) enterprise-grade performance, service delivery and availability. This approach could dramatically lower the total cost of ownership.

SD-WAN achieves this outcome by measuring network traffic metrics (application-aware), such as latency, packet loss, jitter and availability, proactively selecting the optimal path (dynamic link assessment) for each data packet in real-time.

This approach, alongside centralised management for enterprise-wide policy creation/maintenance, makes SD-WAN a natural fit for businesses looking to embrace a multi-cloud strategy. For example, it removes the need for Internet-based traffic to be back-hauled via a central location and/or a data centre.

As with any new technology, the SD-WAN market is highly energised, including multiple startups, as well as the usual suspects such as Cisco, Juniper, etc.

I would position Viptela as my SD-WAN provider, who (in my opinion) have the most mature and reliable service offering, as well as the added benefit of being recently (AUG-2017) acquired by Cisco. I fully expect the SD-WAN market to evolve quickly over the coming years, therefore the Cisco acquisition should help provide stability, as well as complement a Cisco-based LAN/WLAN architecture.

Alongside Viptela, I would target commodity Internet connections as my primary transport for all global sites, positioning CenturyLink for centralised circuit provisioning, support and billing. Through their acquisition of Level 3 Communications, CenturyLink operate a tier-one network, with a strong presence across the Americas and EMA. APAC would likely present an additional challenge, but I am confident that any gaps could be filled through CenturyLink via local provider partnerships.

Alongside the benefits previously described, my decision to position SD-WAN can be summarised based on the following value proposition.

Flexibility: SD-WAN is incredibly flexible, for example, it is easy to increase, reduce and even combine bandwidth across multiple connection types, including fixed-line and cellular. Therefore, it is now perfectly viable to have cellular (4G/5G) as a core part of the WAN architecture, providing an additional level of on-demand resilience for critical sites. Thanks to the centralised management, SD-WAN is also very quick to deploy, without the need for local (on the ground) engagement. This improved flexibility is especially important for modern businesses, that may need to grow (mergers and acquisitions) and contract (divestitures) based on market conditions.

Performance: SD-WAN removes the need to backhaul traffic via a central location and/or data centre, ensuring that traffic routing is optimised across all scenarios, including cloud services (SaaS, Public Cloud, etc.)

Reliability: Thanks to the application-aware routing and dynamic link assessment, it is possible to deliver a highly reliable architecture at low cost, through the use of multiple commodity Internet connections. This approach could dramatically improve service availability, through automated “failover” in the event of a network outage.

Security: SD-WAN inherently includes advanced security capabilities, covering end-to-end traffic encryption and network segmentation. Supported by centralised management, any potential breach and/or vulnerability could be quickly identified and contained (supporting a Zero Trust security model).

Cost: A traditional MPLS architecture could be expensive, driven by a need to procure from a carrier, often resulting in fixed (multi-year) contracts and long provisioning times. It is also common for MPLS networks to have dedicated backup circuits, which only get utilised during an outage. SD-WAN supports multiple connection types (including MPLS), allowing businesses to prioritise inexpensive commodity Internet connections. SD-WAN also aims to optimise traffic across all connections simultaneously, improving efficiency and therefore ensuring a higher return on investment.

In conclusion, I believe the positioning of SD-WAN and the use of commodity Internet connections would unlock a wide range of operational and financial benefits, as well as add flexibility, improving business agility.

macOS and Windows

Thu, 14 Feb 2019 00:00:00 -0600

I recently shared my experience switching from a 15-inch MacBook Pro to a Lenovo ThinkPad X1 Extreme.

In the previous article, I focused on the hardware differences, however, another area of consideration is the software (specifically the “ecosystem” and associated lock-in). For example, if you use a lot of Apple devices and software (e.g. Mac, iPhone, iPad, iMessage, Apple Music) the thought of switching to Windows can be daunting, especially if you have also purchased a lot of third-party software.

As a result, this article will share my journey from macOS to Windows, including the software I use.

It should be noted that as part of my work, I switch between operating systems a lot. On a regular basis, I will use macOS, Windows, Linux, iOS, and Android. Therefore, I have likely been more careful than most in selecting software and services that do not tie me into a specific ecosystem.

With this philosophy in mind, my transition to Windows was fairly painless. Outlined below are the applications I use on a regular basis on macOS, including the Windows equivalent.

As you can see, thanks to the growth of web technologies and frameworks such as Electron, many applications are now cross-platform. Therefore, the transition to Windows can be achieved with minimal disruption or additional cost.

A few minor pain areas include photo editing and vector editing, where I use Pixelmator and Sketch on the Mac. Thankfully, the Serif Affinity suite of software has gained a lot of momentum in recent years, offering a compelling feature-set at a reasonable price-point. My only frustration is that Serif do not offer a transferable license between Mac and Windows (which would likely make me switch on the Mac side).

Software Development can also be a little tricky, specifically when looking to develop native applications for macOS or iOS. The Xcode Integrated Development Environment (IDE) requires macOS and although there are workarounds, most macOS/iOS developers will likely prefer to remain on a Mac. Thankfully, most of my development is web-based, which, thanks to Docker and other supporting build-tools is easily transferred between macOS and Windows.

The most frustrating gap for me is Alfred, which I use on the Mac to launch applications/files, as well as automate my workflow. There have been attempts to “port” Alfred to Windows, specifically Launchy and Wox. However, in my experience, these tools are very immature and lack the polish offered by Alfred (which is bulletproof). Therefore, I find myself using the native Windows search (by hitting the Windows Key). It is a poor comparison to Alfred, lacking any native workflow support, but is at least reliable for basic application/file launching.

Regarding which software ecosystem, I sill favour macOS, where there are a lot of independent, high-quality, low-cost applications such as Alfred, Pixelmator, Sketch. However, over recent years, I feel cross-platform applications have become increasingly impressive, something I expect to continue with Progressive Web Apps (PWA), etc.

In conclusion, I still prefer macOS as an operating system (I am more comfortable using a UNIX-like environment). However, I have been very impressed with Windows 10 (specifically v1809), which has proven to be fast and reliable. Outside of a few notable exceptions (v1803), Microsoft has also done a good job of delivering continuous improvements (including new end-user features), which cannot be said about Apple and macOS (although this might begin to change with the public release of Marzipan).

Tank Arena

Mon, 11 Feb 2019 00:00:00 -0600

In 2017, I wrote a short article regarding GameMaker Studio 2, which I use to develop simple cross-platform, multi-genre video games.

My most recent project is a simple top-down shooter, incorporating “Battle Royale” elements, where the enemies spawn randomly across a battlefield that decreases in size based on a time interval.

The game targets my son, who is only four years old, therefore it must be simple to understand and play, as well short (targeting less than 3mins per play session).

Outlined below are a few high-level requirements:

Minimum of five levels, the final level will continue to spawn enemies until the player tank is destroyed.
Minimum of five enemy types, each with unique abilities, including tank design, strength, and speed.
Each level will increase in difficulty, including different terrain/obstacles, number of enemies, battlefield reduction rate.
Multiple “power-up” items, providing access to new player abilities, including speed, weapons, etc.
Simple scoring system and leaderboard.

Thanks to the simplicity of GameMaker Studio 2, I was able to develop the basic foundations in just a few hours. The video below showcases the first two levels, highlighting the basic movement/shooting mechanics, multiple enemy types, power-up items, scoring and a basic menu system.

At this time, I have not implemented return fire from the enemies, instead, the player is defeated when physically hit by an enemy. I make actually retain this game dynamic for the opening two levels, helping to support a gradual complexity curve for the player.

Thanks to the simplicity and power of the Drop and Drag (DnD) capabilities within GameMaker Studio 2, I often start with DnD, using the “Execute Code” action where required. An example can be seen below, where the basic enemy characteristics have been defined using DnD.

Regarding assets, I often leverage open asset packs as a starting point (my favourite are from Kenney), tweaking the individual assets to meet our personal requirements.

The image below is an example of two enemy tanks, highlighting the collision masks.

The image below depicts the room created for level one, which starts incredibly basic (no obstacles) to help with the learning curve (recognising my target audience).

The next phase is to implement the variable battlefield and leaderboard, followed by the creation of the additional levels and enemies. Thankfully, the additional levels and enemies leverage the existing logic, with new assets and variables (e.g. hit damage, speed, weapon strength, etc.)

Tank Arena can be found on GitHub, all source code and assets are open-source covered under the CC0 1.0 Universal license.

Service Delivery

Sat, 09 Feb 2019 00:00:00 -0600

This article is part of a series. I would recommend reading the articles in order, starting with “Modern IT Ecosystem”, which provides the required framing.

As a brief reminder, this series aims to explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem.

As a technologist, I am eager to start proposing the architecture and associated technologies. However, to ensure the series is credible, I felt it was important to first outline my approach to service delivery.

Introduction

To help ensure quality and consistency, I have defined a service delivery plan, which highlights the key deliverables associated with the implementation of any new service. Depending on the services, the deliverables could be rightsized, helping to promote innovation and agility, except for services that support regulated workloads, where the ability to demonstrate a state of control would be a critical requirement.

For example, as part of this series, I defined a set of business characteristics, where I selected a regulated industry (Food and Drug). Therefore, any service supporting a “GxP” workload would need to adhere to the principles and procedures outlined by Good Automated Manufacturing Practice (GAMP).

Service Delivery Plan

Outlined below is the service delivery plan, which includes the phase, item and decision making authorities (RACI).

Although “Delivery”, “Ops” and “IS” are separate RACI roles, the service delivery plan assumes a “DevSecOps” philosophy. Therefore, security practices should be integrated within the DevOps process, meaning that “Delivery”, “Ops” and “IS” would all be actively engaged as part of the design.

The “Automation” column highlights which items would be viable for automation or facilitate application/data automation. For example, “Application Manifest and/or Infrastructure-as-Code” would be an enabler of future automation.

Where required, the Design Qualification (DQ), Installation Qualification (IQ), Operational Qualification (OQ) and Performance Qualification (PQ) would be embedded as part of the associated phases.

Finally, the service delivery plan is not limited to any specific project management methodology (e.g. Waterfall, Agile, etc.) This would allow for flexibility, depending on the characteristics and/or requirements of the specific service.

Continuous Quality

When operating within a regulated industry, all roles must have an understanding of the associated regulations and guidelines. For example, in the context of “Food and Drug”, it would be important to have a clearly defined process that facilitates Qualification and Validation.

Qualification: The act of proving that equipment or ancillary systems are properly installed, working as designed and comply with the specified requirements.
Validation: A documented objective evidence that provides a high degree of assurance that a specific process consistently produces a product meeting its predetermined specifications.

Qualification is part of the validation, but the individual qualification steps alone do not constitute process validation. For example, the infrastructure must be qualified, while the software running the processes on the infrastructure must be validated.

Many perceive achieving this outcome as a time-consuming process, however, I believe modern technologies and techniques could help streamline the process, whilst simultaneously improving quality.

Continuous Qualification

Software-Defined techniques and the use of Infrastructure-as-Code have a profound impact on the provisioning and maintenance of infrastructure. The four steps outlined below are an example of how these techniques could help to enable and maintain a Qualified State (QS).

As outlined in the service delivery plan, infrastructure requirements must be substantiated as code (Infrastructure-as-Code). This code becomes a blueprint, accurately reflecting the state of the infrastructure.
A “continuous” qualification framework (following standard CI/CD practices), consistently deploys the infrastructure based on a specific blueprint in NON-PRD (DEV, TEST, QA) environments. At which point, automated testing could be performed, resulting in test execution reports.
IT Quality reviews and certifies the blueprints and test execution reports.
The qualified blueprints could be published as a Service Catalogue item, enabling enterprise re-use.

To ensure end-to-end quality, all associated Automation and Monitoring tools must also be qualified, empowering them to proactively deploy and maintain the Qualified State (QS).

Continuous Validation

Similar modern software development techniques could be applied to validation. The three steps outlined below are an example of how continuously validation could be achieved.

Automated testing and test-driven development (TDD) must be a core part of the software development lifecycle.
Functional tests become the backbone of the validation process, where product teams simultaneously write the validation scripts along with the code, executing the tests at regular intervals.
The regular execution of automated tests (before and after implementation) would result in an application that is continuously validated.

A common misconception is that regulators require tests to be executed by a human and physically signed. Using the FDA as an example, the requirement states that “objective evidence that software requirements describe the intended use and that the system meets those needs of the user.” Objective evidence could be achieved through a variety of techniques, including automation.

Conclusion

Regardless of the defined compliance and/or regulatory requirements, I believe every business should aim to proactively embed quality (Quality by Design), making it a continuous part of the service delivery approach. Where viable, software-defined techniques and automation could significantly improve quality, helping to drive higher accuracy and consistency, whilst also unlocking business agility.

ThinkPad X1 Extreme

Wed, 06 Feb 2019 00:00:00 -0600

I first used a Mac in 1996, a Power Macintosh 6000 Series, based on the Quadra 630 form factor. As this hardware pre-dated Mac OS X, I initially favoured my Windows PC, running Windows 95. This changed in 2003 when I purchased a Powerbook G4 running Mac OS X 10.3, which served as my primary computer at university.

Since the launch of the iPhone in 2007, Apple has shifted its focus away from the Mac, looking to maintain its market dominance in mobile computing. As a result, the Mac and macOS have felt a little neglected, with extended product lifecycles and inconsequential updates.

The most significant update in recent years was the release of 2016 MacBook Pro, which included the infamous butterfly mechanism keyboard and Touch Bar. Although this release was a departure from the previous design, the new features were not well received and have since been a point of frustration (Keyboard) and neglect (Touch Bar).

Although I have used a Mac as my daily driver for the past fifteen years, I have also had regular access to Windows PCs, including my custom-built PC and (most recently) a Surface Book 2. However, partly due to my new role, I recently made the switch to a Windows laptop as my daily driver.

I wanted a laptop that would be the equivalent of my 15-inch MacBook Pro, therefore I selected Lenovo ThinkPad X1 Extreme, with the following specification:

Lenovo ThinkPad X1 Extreme
2.6GHz 6-Core Intel Core i7-8850H (Turbo up to 4.30GHz)
32GB DDR4 2666MHz RAM
1TB M.2 NVMe SSD
Intel UHD Graphics 630
NVIDIA GeForce GTX 1050 Ti Max-Q 4GB GDDR5
15.6” 4K UHD Touch IPS Display (3840x2160, 400nits)

This article is not intended to be a detailed review of the ThinkPad X1 Extreme, for that, I would recommend Engadget or MobileTechReview. Instead, I thought I would share my initial thoughts, comparing the primary differences between the ThinkPad X1 Extreme and 15-inch MacBook Pro (specifically the hardware, not the software).

To keep things simple, the article is split into two parts: Positives and Negatives.

ThinkPad X1 Extreme - Positives

Port Selection

The port selection on the Lenovo ThinkPad is exceptional.

2x USB-C Thunderbolt 3
2x USB-A 3.1
1x HDMI 2.0
1x Network Extension Connector
1x SD Card Reader
1x Headphone Jack (Audio/Microphone)
1x Rapid Charge Port

When compared with the MacBook Pro (4x ThunderBolt 3 ports), it is refreshing to be able to use traditional USB devices and connect to a TV/Projector via HDMI, without needing to locate a dongle.

Lenovo has also managed to deliver this wide range of ports, without negatively impacting the size or weight of the laptop. In my opinion, the ThinkPad feels no more cumbersome than the MacBook Pro.

Keyboard

The ThinkPad keyboard is a dream to use, it has excellent key travel (1.7 millimetres) and large, well-spaced keys. Although I have not been able to test over a prolonged period, I feel it is safe to state that ThinkPad keyboard will be more durable than the MacBook Pro, as it leverages a traditional (tried and tested) mechanism.

Although I personally do not mind the butterfly mechanism keyboard used on the MacBook Pro, I do believe the ThinkPad includes a better overall design, allowing for a more comfortable and accurate typing experience.

Display

The 4K UHD Touch Display on the ThinkPad is also a highlight, delivering 186% of the sRGB colour gamut, which is an improvement over the 117% offered by the MacBook Pro. In short, the display is detailed and vivid, making text and photos look great. It also natively supports High Dynamic Range (HDR) for video playback, which is a nice feature, but let down by the poor software support in Windows.

Windows is also not great at scaling to high resolutions, resulting in some occasional visual anomalies such as text blurring. By default, Windows recommends 250% scaling, but personally, I prefer more real-estate, therefore I select 150%. This can make the text quite small but is closer to my experience using the MacBook Pro.

The display is also “touch” compatible, which technically works well, but in my opinion, is fairly useless on a traditional laptop. As the ThinkPad is not a convertible device (does not have a tablet mode), using touch requires you to lean over the keyboard, tapping against the wobbly display hinge. With that said, the display does open flat (unlike the MacBook Pro), which can be useful when working in a confined space.

Performance

Recognising the specification, the ThinkPad is unsurprisingly a great performer. The 6-core (12-thread) processor provides plenty of horsepower, supported by up to 64GB of RAM, a fast NVMe SSD and an NVIDIA GPU.

When compared with the MacBook Pro, Apple has a higher specification CPU (Intel i9), however, it is thermally constrained. As a result, the Intel i7 included in the ThinkPad is actually quicker for real-world tasks. The memory in the ThinkPad is also quicker (2666MHz vs. 24000MHz) and supports up to 64GB. Apple continues to be the market leader when it comes to storage speed (2700MB/s Read/Write), however, as highlighted below, the ThinkPad still delivers strong performance in excess of 2000MB/s Read/Write.

The ThinkPad includes the GeForce GTX 1050 Ti Max-Q, which is fairly old, but still a popular choice as it provides good performance, with respectable power consumption and thermals. Personally, I prefer NVIDIA graphics over AMD, which is what Apple has chosen to include in the MacBook Pro. In my opinion, NVIDIA continues to deliver more innovation and higher overall performance, including proprietary features such as CUDA (parallel computing platform and programming model), G-Sync, PhysX, etc.

The screenshot below highlights the PCMark 10 result, which easily outperformed my Surface Book 2 that scored 3946. As the ThinkPad is not a dedicated gaming laptop, I did not include a 3DMark result, although the NVIDIA GeForce GTX 1050 Ti Max-Q will play most modern games at 1920x1080 (medium settings).

Although not yet confirmed, I suspect Lenovo will announce an updated ThinkPad X1 Extreme in June/July 2019. This will likely include the newer NVIDIA GeForce GTX 1650 Max-Q, offering up to a 20% performance boost.

On paper, the ThinkPad and MacBook Pro are fairly equal, however, as the ThinkPad offers better thermals, it will likely outperform the MacBook Pro when completing CPU/GPU intensive tasks. The added features (specifically CUDA) provided by the NVIDIA GPU are also beneficial when using applications such as Adobe Premier, etc.

ThinkPad X1 Extreme - Negatives

Trackpad

The trackpad on the ThinkPad is my single biggest pain point. In fairness, this is really a failing of Windows laptops, but coming from a MacBook Pro, the Thinkpad trackpad is almost unusable (I now carry a mouse everywhere I go). It is well known that Apple makes the best trackpads, but it is only when you are forced to use a Windows laptop that the difference is fully understood. The MacBook Pro trackpad is incredible, it almost anticipates your movement and never feels sluggish or unresponsive.

On the ThinkPad, I frequently find the trackpad to be unresponsive or difficult to use with any level of accuracy. I have tweaked every possible setting in Windows (including the use of different drivers), hoping to “tune” a better experience, but I have yet to find an outcome that meets my expectations.

Recognising that the trackpad is the primary input mechanism on a laptop, the poor quality offered on the ThinkPad makes using the laptop infuriating.

Build Quality

You might be surprised to see build quality as a negative for the ThinkPad. To be clear, this is a well-built laptop, with a carbon fibre body and soft-touch finish (which is great when typing). Unfortunately, it simply does not compare to the build quality of the MacBook Pro, which (in my opinion) continues to be the best on the market.

As an example, I order a lot of MacBook Pro notebooks at work and can confirm that the level of precision and craftsmanship is always perfectly consistent. I also have access to a lot of ThinkPad laptops, where I have witnessed subtle inconsistencies between identical models.

In summary, the ThinkPad is well built, but I feel lacks the final quality assurance guaranteed by the Apple manufacturing process.

Battery Life

Lenovo state that the ThinkPad will achieve twelve hours of battery life. Unfortunately, I can only assume this result was achieved on the lowest specification (no 4K display), running a very light workload. In my experience, I have been getting closer to five hours of battery life, which can be restrictive.

For example, I travel from the UK to the US a lot and do not always have access to power whilst in the air. With my MacBook Pro, I will easily get nine hours of battery life, running a combination of Atom, Pixelmator, Docker and GameMaker Studio 2 at 50% display brightness. This is simply not possible with the ThinkPad.

On a positive note, the ThinkPad does include a 135W quicker charger (80% battery life in thirty minutes) and can be charged at a slower rate using USB-C. Therefore, assuming you have access to power, the battery life limitation is manageable.

Aesthetics

This is completely superficial and also subjective, however, I would describe the ThinkPad style as understated, maybe even anonymous. In some respects, I like this “no thrills” design approach, as it clearly positions the laptop as a utility, focused on getting work done.

However, I suspect style conscious individuals will find the ThinkPad design to be a little uninspired, lacking any real unique design identity (outside of the iconic TrackPoint button). This is especially true when compared against the MacBook Pro and even PC competitors such as the Dell XPS line.

ThinkPad X1 Extreme - Conclusion

In summary, the ThinkPad X1 Extreme is a great laptop! It is fast, functional and (in the majority of cases) a pleasure to use.

Likely helped by the fact that I still use Windows on a regular basis, but I was pleased with how quickly I was able to adapt to using the ThinkPad as my daily driver. There are even areas I prefer over my MacBook Pro, specifically the keyboard and access to an NVIDIA GPU.

Unfortunately, as crazy as it sounds, the trackpad is a potential deal-breaker for me. Although I use a lot of keyboard shortcuts, I still rely heavily on the trackpad, therefore, to have such an important part of the laptop be this unreliable is unacceptable. I appreciate the ThinkPad includes the TrackPoint and touch display, however, in my opinion, these are simply not a viable replacement for a good trackpad. Therefore, if I end up switching back to a MacBook Pro as my daily driver, I suspect it will be the trackpad that pushes me over the edge.

Finally, it is worth highlighting the price of ThinkPad when compared to an equivalent MacBook Pro.

ThinkPad X1 Extreme (6-Core, 32GB, 1TB, 4GB GPU) = £2773.19
15-inch MacBook Pro (6-Core, 32GB, 1TB, 4GB GPU) = £4004.00

It is difficult to ignore that price difference. A saving of £1230.81 is a lot to pay for a better trackpad!

Modern IT Ecosystem

Mon, 21 Jan 2019 00:00:00 -0600

The desire to modernise an enterprise business is often a familiar battle with legacy and technical debt. This article is the first in a series, which will explore the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up.

Over the coming months, I will share my thoughts regarding the creation of a modern IT ecosystem. I will take on the persona of an enterprise architect, therefore the articles will predominantly focus on the architecture, technologies and positioning, however, I will also highlight key processes and methodologies.

Recognising that all good IT decisions must include business context, with a clear value proposition. I have defined the following hypothetical business characteristics.

Global Business (Americas, EMA, APAC)
Regulated Industry (Food and Drug)
Customers in 90+ Countries
Value Chain includes R&D, Manufacturing, Commercial and Global Services
15,000+ Total Users (Employees and Contractors)
170+ IT Users (Employees and Contractors)
50+ Physical Sites Globally
10+ R&D and Manufacturing Sites (24x7 Operation)
Established Traditional Business Model
Emerging Digital Business Model
$5Billion Revenue

Selecting a regulated industry (Food and Drug) enforces a set of rules, regulations and guidelines. This decision will impact the design of the ecosystem, but also hopefully add credibility to the analysis, recognising that the proposed architecture must meet a minimum set of standards, governing safety, efficacy, quality, privacy and security.

Throughout this series, I will use these business characteristics to inform my proposed architecture, as well as take into consideration other key business drivers such as maximising the return on investment.

IT Principles

Prior to any architecture decisions, I believe it is important to baseline the approach by defining a set of IT principles. These IT principles aim to drive better strategic decision making, by proactively triggering the right architecture conversations.

I have established four principles assigned to the “classic” enterprise architecture domains, covering Business Process, Information, Application and Technology.

The rationale behind having only four principles is to ensure that they could be easily consumed and remembered. As highlighted in my 2016 article “Enterprise IT Architecture”, architects tend to overanalyse, resulting in unnecessary complexity which could limit adoption and adherence.

IT Declarations

Although the IT principles provide a high-level direction, they could be difficult to translate into a tangible outcome. As a result, I have created a corresponding set of IT declarations, which aim to add additional context, reinforcing the target-state architecture.

Hybrid-Cloud: By default, Public Cloud must be prioritised over Colocation Data Centres and Edge Computing.
Open Source: By default, new applications must prioritise open-source technologies, with any proprietary technologies looking to maximise strategic investments.
Custom Developed: By default, new custom-developed applications must embrace a Cloud Native architecture (Twelve-Factor App and FAIR), prioritising FaaS and PaaS.
Commercial of the Shelf (CotS): By default, new Commercial of the Shelf (CotS) applications must embrace a Cloud Native architecture, prioritising SaaS.
Server Deployment: By default, new IaaS deployments must target immutable infrastructure, including ephemeral (short-lived) and stateless (do not persist a session) servers.
Automation: By default, new application deployments must be automated, leveraging build tools and/or automation services.
Datastores: By default, new managed datastores must prioritise open-source technologies, avoiding proprietary/licensed technologies.
Integrations: By default, new system-to-system integrations must leverage the approved Enterprise Integration Fabric.
API-First: By default, new applications must leverage open and documented web service APIs and/or Webhooks.
Client Plugins: By default, new client-side applications must be browser-based, avoiding the use of proprietary plugins (e.g. .Net Framework, Java, Flash, etc.)
Internet Accessible: By default, new applications must be accessible via the Internet, secured at source and in transit, removing the need for a VPN/VDI.

I would not anticipate 100% adherence to the IT principles and/or IT declarations. For example, specialised capabilities from R&D and Manufacturing, which commonly require proprietary software/hardware.

Modern IT Ecosystem

With the previously stated business characteristics in mind, the diagram below highlights my vision of a modern IT ecosystem. It is purposely “marchitecture” (a combination of marketing and architecture), used to communicate the desired end-state to a non-technical audience.

The bullets below provide context to the diagram, describing the philosophy and intended direction for each area. As part of this series, I plan to analyse these areas (among others) in greater detail.

SaaS: Recognising the limited competitive advantage, all highly-industrialised/commodity capabilities would be positioned as SaaS. Examples include Productivity, Collaboration, Service Management, Customer Relationship Management (CRM), Human Resource Planning (HRP), etc.
Public Cloud: Public Cloud would be prioritised for all application/data hosting, emphasising “up the stack” services. For example, FaaS and PaaS would be prioritised over IaaS. The underlying architecture would strive to be cloud-agnostic, embracing a hybrid multi-cloud strategy.
Colocation Data Centres: Multiple Colocation Data Centres would provide support for capabilities that would not be commercially or architecturally a good fit for Public Cloud. For example, the previously mentioned, specialised capabilities from R&D and Manufacturing. The Colocation Data Centres would securely and dynamically connect to each other via a managed network backbone, which would also provide high-performance connectivity with key partners and Public Cloud providers.
SDDC: The infrastructure deployed within the Colocation Data Centres would be highly-converged, leveraging modular (blade-based) hardware that would allow compute, storage and network to be scaled independently (providing flexibility). All Data Centre resources and services would be software-defined, following modern Software-Defined Data Centre (SDDC) techniques.
ERP: Enterprise Resource Planning (ERP) would be fully managed (SaaS-like), but hosted in a specialised cloud that would ideally interconnect with managed network backbone provided by the Colocation Data Centres. Recognising the importance of ERP, this approach would aim to exploit the benefits of a fully managed service, whilst offering “LAN-like” performance, reliability and consistency. It would also help avoid unnecessary lock-in with a specific Public Cloud provider (AKA Hyperscaler), which could become a costly investment based on ingress/egress charges.
IAM: Identity Access Management (IAM) would support the Cloud (SaaS, Public Cloud), Colocation Data Centres and Edge Computing, providing flexibility and resilience for any application. Modern authentication and authorisation mechanisms would be prioritised, enabling single sign-on and password-less capabilities for users.
Automation: All cloud services (SaaS, Public Cloud), Colocation Data Centre and Edge Computing would be fully automated, leveraging Software-Defined techniques. As depicted by the “DevOps” cloud, the automation would be delivered via a unified suite of services, covering provisioning, governance, budgeting, discovery, testing, security analysis, quality assurance, documentation and change control.
WAN: The Wide Area Network (WAN) would utilise an SD-WAN architecture, simplifying management and operations, whilst enabling the use of commodity Internet circuits instead of traditional MPLS.
User Endpoints: Applications and services would be Internet-accessible, via the web and/or an API. This approach would simplify the endpoints architecture, removing the need for specialist hardware and/or software, whilst promoting a “consumer-like” experience covering any operating system (e.g. Windows, macOS, Linux, ChromeOS, iOS, Android, etc.)
LAN: Wireless technologies would be prioritised at the local sites, made available via a unified SSID that would dynamically provision access based on the endpoint conditions. R&D and Manufacturing would include strict segmentation policies, following the Purdue Reference Architecture, allowing for a clear separation of concerns and ensuring sensitive workloads could run autonomously.
Edge Computing: To support latency-sensitive applications at R&D and Manufacturing sites, edge compute and storage would be made available. Similar to the Colocation Data Centre architecture, Edge Computing would target a highly-converged (possibly hyper-converged) infrastructure, supported by software-defined techniques.
Information Security: A Zero Trust security model would be followed, localising and isolating threats through microcore, segmentation and deep visibility. To achieve this outcome, multiple technologies would be embedded as part of the wider architecture, including Layer 7 Firewalls, Network Segmentation, Agent-less Network Access Control, Data Loss Prevention, Privileged Access Management and a dedicated Security Operations Centre (SOC). These technologies would be complemented by a clear “least privilege access” strategy and appropriate governance, monitoring, auditing, etc.

Conclusion

Over the coming months, I plan to build upon this initial article, incrementally detailing the capabilities associated with my vision of a modern IT ecosystem. As highlighted in the framing, these articles will reference the previously defined business characteristics, predominantly focused on the architecture, technologies and positioning.

Career Planning

Sat, 12 Jan 2019 00:00:00 -0600

In September, I transitioned from my role as an Enterprise IT Architect to Chief Technology Officer (CTO). This change marked a major milestone in my career, something I had been working towards for over a decade.

This article highlights my approach to career planning, covering the high-level methodology and some key techniques.

As a general rule, embracing stoic philosophy, I attempt to identify, prioritise, and own the choices that are clearly within my direct control, whilst accepting external events that are beyond my direct control.

This mindset has been important throughout my career, helping me to prioritise actions that can contribute to a positive outcome, disregarding external factors that are impossible to change or avoid.

It should be noted that this approach can be perceived as lacking empathy, as acceptance, without context, can appear “heartless”. However, I believe focus and discipline are key behaviours that promote success, recognising that external events are a common distraction that can derail progression towards a specific goal.

With this in mind, I consider my career a strategy, which must be effectively planned and methodically executed. To help, I have defined a simple five-step career planning methodology.

Mission Statement
North Star
Career Plan
Education Plan
Community Network

This article will cover each step, providing examples from my career.

It is important to note that every career plan is unique to the individual and the methodology described in this article compliments my style and therefore may not resonate with everyone.

Mission Statement

A mission statement is a formal summary of the aims and values of a company, organisation, or individual.

As with any strategy, I believe it is critical to establish a clear outcome. In the context of career planning, a mission statement can help establish your personal brand, whilst communicating your purpose.

For example, my mission statement reads:

Apply technology to improve the wellbeing of people and our planet.

There is no right or wrong answer when defining a mission statement. However, I believe it should focus on an area of passion, which will encourage resilience and perseverance during times of stress and/or hardship. For example, my mission statement attempts to articulate three points:

I am passionate about “technology”.
I am passionate about how to “apply” technology, not invent/create it.
I am passionate about health and “wellbeing”, specifically people and our planet.

I use my mission statement as an anchor to ground my career, ensuring I stay true to my passion area and purpose.

North Star

With a mission statement established, identifying a position/role that best supports the outcomes of the mission statement can help solidify the direction.

For me, that would be a position/role that includes the highest level of influence and accountability over key decisions, however, this might not be desired by everyone. For example, not everyone desires a leadership position, which is perfectly acceptable.

It was this reflection, over a decade ago, that led me to the role of Chief Technology Officer (CTO), ideally, as part of an organisation focused on the wellbeing of people and our planet.

In addition to the position/role, establishing a set of principles can help provide scope to your career plan, serving as guardrails or system of belief (e.g. morals, ethics, etc.)

At a minimum, these principles should cover five key areas.

Self (Physical/Mental Health)
Family (Parents, Children, etc.)
Friends
Social (Work/Life Balance)
Financials (Commitments, Protections, etc.)

The principles should describe ideals, which can be malleable and (in certain scenarios) even compromised, but should stay ultimately true for sustained personal wellbeing. For example, if you have goals related to family, these must be protected/balanced alongside your career.

Career Plan

Once a north star (position/role) and supporting principles have been defined, I document the expertise, experience and behaviours required by the target position/role. This can be subjective, especially if the position/role is not well defined, but still an important exercise to help plot a path forward.

The list outlined below includes a few example behaviours that I deem critical to be a successful Chief Technology Officer (CTO).

Technical Depth and Breadth
Business Context and Credibility
Strategic Mindset
Communications and Influence
Proven Record of Enterprise Delivery
Quality External Network

Once understood, you can start to chart a career plan, which could include multiple paths, roles, and organisations. It is also worth including any previous experience, as this can help highlight gaps and inform future decisions.

The goal is to construct a plan that incorporates opportunities that enable you to learn and demonstrate mastery of the previously defined expertise, experience and behaviours, always pushing towards your desired outcome (north star). The diagram below is an example of my career plan.

As you can see from the example, my career plan includes multiple milestones, which come in the form of new roles, short-term assignments and/or opportunities. Ideally, each should have a specific purpose, tangibly contributing towards your personal development.

To help support healthy progression throughout the career plan, I follow a simple methodology that I affectionately call “The Bourne Engagement”.

When starting something new, either a role, project, or a programme, I follow a three-phase engagement approach, which coincidently matches the titles of the popular Bourne series of movies. For example.

Starting with “Identity”, I believe it is critical, when starting something new, to spend time establishing a strong relationship with the team, whilst seeking to understand any relevant business context, and/or business politics/bureaucracy.

In theory, this approach helps to form a strong foundation, established in trust, whilst also ensuring the team has a clear understanding of your personal brand, areas of passion, etc.

Only once your identity has been established, you can transition to “Supremacy”, which is where you demonstrate your strengths and influence, seizing opportunities and positioning yourself as a high performer and thought leader.

Finally, with an established identity and sustained/consistent high-performance, you can transition to “Ultimatum”, where you prepare for career progression, through clear accountability and by leveraging your established credibility and people network.

Although slightly “cringe-worthy”, this three-phase approach has always served me well, regardless of situation or level. The speed at which you can transition through each phase is dependent on the specific scenario, but I would caution attempting to proceed too quickly, as trust and credibility can not be cheated, it requires “time in the trenches”.

Education Plan

In parallel to the career path, I also produce an education plan. It is easy to assume that formal education concludes at the end of school/college/university, but I believe it must be seen as a continuous process, especially when working in a fast-moving industry such as IT.

As with the career plan, the education plan should connect to the previously defined required expertise, experience and behaviours. It can be used to enrich and accelerate career progression, whilst also bringing a valuable “external perspective”, ensuring diversity of thought, which can be lost when working within a single organisation for a prolonged period.

Outlined below is an example of my education plan, which includes multiple channels, covering formal education, courses, conferences, speaking opportunities, etc.

Alongside the value of continuous learning, the education plan can help establish a strong external people network, which can act as a differentiator as you progress through your career.

I have found speaking at (not just attending) events (e.g. conferences, meet-ups, etc.) to be an excellent way to engage the community and establish external relationships.

Community Network

The final area formalises the importance of the people network. As it relates to career, there is certainly truth to the common saying “it’s not what you know, but who you know”, however, I would argue both are equally important.

With that said, a strong people network can help ensure your personal development is diverse and inclusive, whilst also revealing new opportunities that might otherwise be hidden. Therefore, I take time to establish and nurture a robust people network, supported by a simple stakeholder map.

The goal is to identify key individuals (internal to your organisation and external) that can support your personal development, aligned with your career plan and defined expertise, experience and behaviours. A stakeholder network can be quite large, which is why categorisation is useful, as it provides a simple framework to help coordinate engagement (type of engagement, frequency, etc.)

In parallel, key individuals should be positioned as formal mentors. A strong, trusted mentor can have a profound impact on your career, however, it is the mentee that must drive this value. For example, I identify mentors that can directly support my personal development, focusing the conversations on specific topics vs. generic chat.

As you progress in your career, the concept of a “reverse-mentor” becomes equally important, targetting an early career professional who can help provide a generational perspective, alongside future trends that may not be obvious in leadership positions.

In short, I consider a strong people network a critical asset that must be proactively maintained, with specific goals, to ensure the value is maximised.

Conclusion

As stated during the framing of this article, every career plan is unique to the individual and the methodology described in this article may not resonate with everyone.

Regardless of your approach, one thing is true, you must own your career! Therefore, be proactive, get organised and start working towards your personal outcome!

Thriva

Thu, 03 Jan 2019 00:00:00 -0600

My two primary passions in life are Technology and Health, therefore I am always interested when these two areas converge.

Over the years, I have written a number of articles focused on improving health outcomes through the use of technology (see below). I have also experimented with new concepts and services, such as the Quantified Self, 23andMe, and Babylon Health.

Thriva, which launched in 2016, has always been on my radar. The service aims to take health tracking to the next level, through a regular blood test which can be completed at home. Although this mission sounds similar to the ill-fated Theranos, Thriva targets a limited set of tests, enabled through traditional lab testing techniques.

I like the concept a lot, as it shifts the focus from reactive to proactive (preventative) health, which is an area I have always promoted. As highlighted on their website, Thriva state:

“By increasing convenience, ease, and insightfulness, we’re providing the products and services that will empower people to proactively manage their health. Once you know what’s going on inside your body, you can understand the impact of your lifestyle on your health and start to make small changes that add up to a healthier you.”

My Experience

The ordering process is very easy, simply sign-up for an account on the Thriva website and select your subscription level, covering Essential, Baseline, and Advanced. I selected “Advanced”, which costs £69.00 and includes the following tests:

Cholesterol
Liver Function
Iron Profile
Vitamin D
Vitamin B12
Vitamin B9 (Folate)
Diabetes (HbA1c)

I was a little suspicious about how easy the process would be to collect the blood, especially the size of the sample (which is not clearly highlighted on the homepage). It does however state:

“Your kit contains everything you need for your simple finger-prick blood test.” and “The whole process is straightforward and takes just a few minutes.”

The sample collection box arrived in the post very quickly, which is compact and therefore should fit through a standard letterbox.

Upon opening, the box is clearly marked, with focus placed on the instructions and three phases (prepare, sample, send).

The sample collection box included:

2x Sample Collection Tube
1x Alcohol Swab
1x Moist Wipe
2x Plasters
3x Spring Loaded Lancets
1x Plastic Protective Casing
1x Return Form
1x Return Envelope

This is where I encountered my first challenge, as the website and instructions state that there should be only one Sample Collection Tube, but I had received two. The Sample Collection Tubes were also a little larger than I had anticipated, especially as the only way to fill them is to drip blood from a finger-prick created by the lancet.

Even with the confusion, I decided to start the process, with the aim to fill both tubes. The step-by-step instructions were reasonably clear and I followed them to the letter.

Unfortunately, no matter how many times I pricked my finger, I simply could not get a constant flow of blood. After 15 minutes of trying, the prospect of filling two tubes had become daunting.

Eventually, I decided to give up, the process (in my experience) had not lived up to the “simple” and “straightforward” marketing.

My next step was to contact Thriva via the FAQ section on their website. This led me to the article “I’m having trouble collecting a blood sample”, which states the following on the opening line:

“Collecting your sample for the first time can be tricky.”

This was the first time the process had been described as “tricky”, which in my opinion makes the homepage statements a little misleading.

Thankfully, the knowledge article itself was useful, clearly stating that Thriva will provide a free replacement kit, in the event the first attempt fails to achieve a usable sample. I can also confirm that if you decide not to proceed, Thriva will offer a refund (excluding a £7 fee for the kit itself).

At this time I have not decided if I plan to re-try the process, questioning if the benefit is worth the effort.

Conclusion

In conclusion, I remain excited by the prospect of a service that enables individuals to proactively manage their health, especially with the high level of accuracy that can be achieved through a blood sample.

However, in my experience, the Thriva process is uncomfortable and cumbersome and the statement “simple finger-prick blood test” does not accurately describe the multi-minute blood dripping exercise that is required to fill the two tubes. I believe this process would be enough to deter most people, especially for anyone who is uncomfortable around blood.

Due to the complexity of the process, I see Thriva as a service that would only be accepted by “health enthusiasts”, not the general public. This is a problem for Thriva, as their mission is only achievable if the data provided by their service can produce actionable insights for the demographic most at risk (e.g. General Public).

With that said, every revolution must start somewhere and I appreciate the pragmatic approach taken by Thriva, delivering a service that places a focus on accuracy (leveraging mature techniques and accredited labs), even if the sample collection process is cumbersome. Hopefully, through technology advances, they can continue to simplify the process and therefore attract a wider audience.

Amazon Go

Mon, 12 Nov 2018 00:00:00 -0600

I was recently in Seattle meeting Microsoft but took the opportunity to visit an Amazon Go (7th & Blanchard).

Amazon Go is an automated grocery store, allowing you to purchase items without any human intervention. You simply enter the store by scanning a QR code on the Amazon Go app. Once inside, you are free to fill your bag with items and leave when finished.

The two images below are from my smartphone, showing the key (I have removed the QR code) and receipt for my purchase. The receipt appears within the app after a few minutes of leaving the store.

Unsurprisingly, the store is packed with technology to support the purchase, checkout, and payment processes. This includes computer vision, deep learning, and sensors, specifically a camera array on the ceiling (see image below) and weight sensors under specific items.

The combination of these technologies allows for each customer to be tracked whilst in the store, including what items they take (and return) from the shelves.

I cannot speak to the accuracy of the process, but it did work flawlessly for my purchase. This included multiple people entering the store under my QR code, each removing and returning multiple items on different shelves.

Overall, I was impressed with the process and admire Amazon for trialing a radically different and innovative approach to grocery shopping. As a customer, it was certainly a strange experience, entering and exiting a store without any obvious payment process.

I do have questions about the scalability of the approach, as I assume the technology (today) is far more expensive than simply having a couple of staff support the checkout process. It also has specific limitations, for example, due to local laws, human intervention is still required when purchasing alcohol, etc.

With that said, I hope Amazon continues to refine the process and opens more stores around the world.

Samsung Q7FN

Sat, 10 Nov 2018 00:00:00 -0600

In 2016, I shared my entertainment setup, including my TV, Sound, AV Receiver, and Universal Remote.

I recently upgraded my TV to a Samsung Q7FN QLED. The full specification can be found below:

Samsung Q7FN QLED (55-inch)
3840×2160 Resolution
240Hz Refresh Rate (120Hz Measured)
1500nits Peak Brightness
10bit Colour Support
HDR 10+ Compatible
100% DCI-P3 Coverage
4k Variable Refresh Rate (48Hz to 60Hz)
1080p Variable Refresh Rate (20Hz to 120Hz)
12.6ms Response Time
Edge LED Backlight

As highlighted in my previous article, I wall mount my TV, with all AV equipment located in a cupboard under the stairs. Thankfully, the Samsung Q7FN is compatible with my existing VESA wall mount, therefore the upgrade was fairly painless.

I often get asked, why did I not purchase an OLED?

My usage pattern includes sport, games, and music. Each of these includes static content, which can often remain fixed on the display for many hours at a time (especially music, when streaming from Spotify). OLED is the gold standard when it comes to picture quality, however, it does suffer from burn-in when displaying static content for prolonged periods.

If you are interested in learning more about OLED burn-in, Rtings have completed a very comprehensive test across six LG TVs and 5000+ hours. I have also captured their recommendation below:

“Those who display the same static content over long periods of time should consider the risk of burn-in (such as those who watch lots of news, use the TV as a PC monitor, or play the same game with a bright static HUD). Those who are concerned about the risk of burn-in should go with an LCD TV for the peace of mind.” - Rtings

Another reason for selecting a Samsung is the inclusion of the One Connect box, which enables a very clean and minimal setup. The Samsung Q7FN comes with an upgraded One Connect box, which not only includes all input/output connections but also power. Therefore, I simply need to route one (very thin) cable through the wall to my AV equipment.

It is worth noting that although the Samsung Q8FN includes local dimming (40 zones), it does not support the One Connect box, therefore my purchasing decision was between the Q7FN (RRP £1509) and Q9FN (RRP £2119). I decided to go with the Q7FN as I expect I will upgrade again within a few years, however, if you want the very best on offer from Samsung, you should target the Q9FN.

Overall, I have been very impressed with the Samsung Q7FN, specifically the improved HDR quality, as well as advanced gaming features such as Variable Refresh Rate (FreeSync 2). The Edge LED Backlight still results in some blooming, but overall black uniformity is very good (especially when compared against my old Samsung KS8000).

As with any new TV, to achieve the best results it must first be calibrated. As an example, Tom Cruise and Christopher McQuarrie recently released a video (embedded below) highlighting the impact of motion interpolation (soap opera effect), which is often enabled by default and can distort the content.

Outlined below are my calibration settings.

NOTE: I would always recommend professional TV calibration, however, this requires specific tools and a detailed understanding of colour space, etc. As a result, the calibration settings outlined below should be used as a reference only.

Samsung Q7FN Calibration

Eco Solution = Off
Picture Mode = Movie
Backlight = 25
Brightness = 0
Contrast = 45
Sharpness = 0
Color = 25
Tint (G/R) = G0/R0
Digital Clean View = Off
Contrast Enhancer = Off
Local Dimming = High
Auto Motion Plus = Custom
Blur Reduction = 0
Judder Reduction = 0
LED Clear Motion = Off
Color Tone = Warm2
Gamma = +2
Color Space = Auto
HDMI UHD Color = On (All Sources)

The settings outlined above are configured for all modes (including HDR, Gaming, etc.)

However, when displaying HDR content, I ensure the “backlight” is set to maximum, allowing for the highest possible peak brightness (1000-4000 cd/m²). When gaming, “Game Mode” should be enabled to ensure the lowest possible input latency (12.6ms).

GeForce RTX vs. 3DMark

Sat, 03 Nov 2018 00:00:00 -0500

3DMark is a popular computer benchmarking tool targeting 3D graphics and CPU performance. It produces a normalised score (higher is better) for comparing different PC hardware configurations.

I joined the 3DMark community (known then as MadOnion) in 2000, just after the launch of 3DMark2000. My small claim to fame, in 2002, I gained early access to an NVIDIA GeForce Ti 4600, which I modified with a custom cooler and voltmod. This resulted in a short period near the top of the 3DMark2001 SE global leaderboard.

Earlier this month, I gained access to an NVIDIA GeForce RTX 2080 Ti. As part of my testing, I thought I would take a trip down memory lane, running all compatible versions of 3DMark.

Introduction

As outlined in the article “my setup”, the GeForce RTX 2080 Ti was added to my custom-built PC. The full specification can be found below:

Asus Maximus VIII Hero Alpha (Intel Z170)
Intel Core i7-6700K 4.0GHz @ 4.6GHz
Noctua NH-D15 CPU Cooler
Corsair Vengeance LPX 32GB DDR4 PC4-24000C15 (15-17-17-35)
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
Samsung 950 Pro 512GB M.2 PCI-e NVMe SSD
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case

NOTE: The GeForce RTX 2080 TI is obviously not optimised to run legacy benchmarks. Therefore, these results are for reference only and should not be used to compare hardware.

3DMark (2013 to 2018)

3DMark is a DirectX 12 benchmark for Windows 10, supporting multi-core processors and 4K resolution. 3DMark is the only fully supported benchmark, therefore the result below can be used for comparison.

3DMark will be updated in January 2019 to include a ray-tracing benchmark, allowing for more thorough testing of the GeForce RTX series. This new suite of tests will become increasingly important as ray tracing and hybrid rendering (ray tracing + rasterisation) matures.

3DMark 11

3DMark 11 is a DirectX 11 benchmark, including features such as tessellation, compute shaders and multi-threading.

As a relatively modern benchmark, the GeForce RTX 2080 Ti produces an impressive result. By comparison, in 2016 my GeForce GTX 980 achieved a score of P18063.

3DMark Vantage

3DMark Vantage is a DirectX 10 benchmark for Windows Vista. It includes the Jane Nash and New Calico graphics tests, an AI test, a physics test and six feature tests.

The Jane Nash test achieved an average of 201 FPS, followed by 308 FPS for the New Calico test.

3DMark06

3DMark06 is a DirectX 9.0c benchmark, adding support for Shader Model 3.0 and new CPU tests using PhysX. The graphics tests are Return to Proxycon, Firefly Forest, Canyon Flight, and Deep Freeze.

The GeForce RTX made short work of the game tests, hitting an average of 128 FPS in Return to Proxycon and 275 FPS in Canyon Flight. Even though the benchmark is over a decade old, the CPU tests remain incredibly intensive, with my quad-core/eight-thread Intel i7 only achieving an average of 5 FPS.

3DMark05

3DMark05 is a DirectX 9.0c benchmark, providing support for Shader Model 2.0. Game tests include Return to Proxycon, Firefly Forest and Canyon Flight.

The Return to Proxycon test achieved an average of 146 FPS, with Canyon Flight averaging 320 FPS. Interestingly, the CPU tests in 3DMark05 performed worse than 3DMark06, achieving an average of 1.3 FPS.

3DMark03

3DMark03 is a DirectX 9 benchmark. It was the first 3DMark to include CPU tests. The graphics tests are Wings of Fury, Battle of Proxycon, Troll’s Lair and Mother Nature.

It was impressive to see Battle of Proxycon hit an average of 1462 FPS and the Pixel Shader 2.0 performance exceed 3300 FPS. Based on the numbers, 3DMark03 game tests resulted in the highest overall performance.

3DMark2001 SE

3DMark2001 SE is a DirectX 8.1 benchmark. The Car Chase, Dragothic, Lobby, and Nature tests feature vertex and pixel shaders and full scene anti-aliasing.

The user interface of 3DMark2001 SE is restricted to 999 FPS, therefore it is more difficult to understand the true performance. However, using FRAPS revealed and an average of 1300 FPS for the opening three (low detail) graphics tests.

Conclusion

I was also able to run 3DMark2000 (DirectX 7.0), however, this required “Windows XP Compatibility Mode” to be enabled, as well as a modified “SystemInfo.dll”. Although the benchmark did run clean, the score was drastically low (4295), clearly highlighting the impact of unoptimised hardware/software.

As a long-time user of 3DMark, it was fun to watch the GeForce RTX 2080 Ti tackle these legacy benchmarks. It was also surprising how well all of the benchmarks performed, considering the lack of support and optimisation.

UK EA Forum - Part VI

Fri, 02 Nov 2018 00:00:00 -0500

In 2017, I launched a UK Enterprise Architecture Forum, which supports an open community of architects, allowing them to share ideas, experiences, and challenges.

The forum aims to meet once per quarter, however, we also collaborate throughout the year via digital channels. Details of the previous meetings can be found linked below:

Today, we held the sixth forum, where we covered the following topics:

IT4IT - IT Service Management
Architect Skills / Behaviours
Enterprise Architecture Principles
Greenfield Opportunity

Outlined below are my notes from the meeting.

IT4IT - IT Service Management

IT Service Management has always been a challenge for businesses with a decentralised operating model.

Technologies such as ServiceNow can help, through the use of self-service, workflows, approvals, etc. However, the service is only as good as the business process, which is not inherently defined.

BP shared an update regarding their digital modernisation initiative, which included a new framework to support IT Service Management. The easiest way to understand the framework is to watch the session they recently delivered at the Now Summit in London.

The framework de-emphasises centralised IT, by embracing an “Agile-like” approach, where any team (decentralised) can create a service and make it available for consumption via an internal marketplace. A key ramification of this change is accountability, which now sits with the team that created the service (including lifecycle, change management, security, quality, privacy, etc.)

The framework is split into three parts.

A Service (A group of resources).
A Service Offering (What people consume).
A Resource (What makes up the service).

To ensure the framework can scale naturally, with minimal governance, the following constraints are applied.

A resource must be in one and only one service.
No direct relationship between a resource and a product.
Users inherit roles only from the groups they are in.

The goal for BP is to enable an internal marketplace (an API ecosystem), where teams can create, share, reuse and update services, without manual intervention or unnecessary bureaucracy. In a perfect world, these services become highly automated (improving agility), easy to manage (clear dependency mapping), as well as continuously maintained (clear accountability).

The framework does not replace the need for an overarching governance. For example, Enterprise Architecture must still help to define strategic technologies, positioning, etc. It is also not a “silver bullet”, meaning that it might not be the right fit for every company (depending on the size, organisational structure, culture, etc.)

BP is looking to build an open community around the framework, therefore, if you are in the process of enabling a new IT Service Management capability, I would recommend connecting via our LinkedIn Group.

Architect Skills / Behaviours

The forum discussed the evolving role, skills, and behaviours of an architect. The following quote was highlighted from “Accelerate” (Forsgren, Humble, Kim).

*High performance is possible with all kinds of systems, provided that the systems - and the teams that build and maintain - them are loosely coupled.

Testability and deployability are important characteristics.

Architects should focus on engineers and outcomes, not tools and technologies.*

Everyone agreed that Enterprise Architecture must drive a clear value proposition and act as an enabler (not a barrier or an inconvenience). To achieve this outcome, architects must be an embedded part of the business process, supporting the design and delivery of IT solutions to meet business outcomes.

I shared the following behaviours, which I have used to help set clear expectations across the architecture community at my company. It is worth noting that not all of these expectations will apply to everyone business.

Business Outcomes

All architecture decisions must enable the business, with a clear purpose and value proposition.
Best in class technology, may not be best for our business. Focus on the business requirements, not the technology.
Maximise investments, delivering IT solutions at the highest quality, with lowest cost and risk.

Accountable

Industry analysts and vendors can provide valuable insight, but you own the decision.
Include business continuity into planning and decision-making.

Inclusive

Ensure all architecture decisions support the broader strategy, with a clear understanding of all dependencies. This must include security, quality, and privacy by design, following a risk-based approach.
Consider global and regional requirements, specifically related to security, privacy, and regulatory constraints.
Focus on relationships, partnering to translate technology into business outcomes.

Community

Challenge each other with respect and always assume good intentions.
Be an enabler, not a barrier. Directly contribute to enabling positive outcomes through collaboration.
Leverage the community, thinking reuse before buy, buy before build.

In many respects, the role of an architect is no longer a dedicated discipline. Instead, like Security, Quality, and Privacy, everyone must be accountable for architecture. With this in mind, the evolving role of an architect is to help enable this mindset.

Enterprise Architecture Principles

Unsurprisingly, Enterprise Architecture is a common topic at the forum. Generally speaking, most forum members follow their own methodology, taking the best parts from established frameworks such as TOGAF, etc.

I have previously shared my work regarding Enterprise IT Architecture, including Principles, Positioning and Patterns.

This agenda topic focused specifically on principles. I shared the following slide, summarising principles across four domains (Business Process, Information, Application, and Technology).

Although every business is unique, the continued impact of digitisation has resulted in a lot of common themes (e.g. API-Centric, etc.)

Greenfield Opportunity

As a thought experiment, the group discussed the “art of the possible” if an enterprise business could hypothetically rebuild IT from the ground up, creating a modern IT ecosystem. As part of this discussion, I took the opportunity (using the slide below) to share my vision of a modern IT ecosystem.

This topic highlights the power of the UK Enterprise Architecture Forum, where I was able to position a topic for discussion, without fear of bias, and leverage the collective expertise and experience of the group.

Conclusion

If you are interested in the joining the UK Enterprise Architecture Forum, please request access via our LinkedIn Group. All architects are welcome, covering start-up and enterprise businesses.

GeForce RTX

Fri, 26 Oct 2018 00:00:00 -0500

On 20th August, NVIDIA revealed their next generation consumer graphics hardware, known as the GeForce RTX 20 series.

The series launched with three consumer graphics cards, the GeForce RTX 2070, GeForce RTX 2080, and the flagship GeForce RTX 2080 Ti. These cards started to hit the market in September, but are likely to remain in short supply throughout 2018.

The GeForce RTX 20 series introduced the Turing architecture, fabricated by TSMC on their 12nm FFN (FinFET NVIDIA) process. The series will eventually supersede the current Pascal architecture found in the popular GeForce GTX 10 series. Therefore, if you are interested in a GeForce GTX 10 series graphics card (e.g. Geforce GTX 1080), now would be a good time to purchase.

NVIDIA state that the Turing architecture represents the biggest technical leap in over a decade, enabling major advances in efficiency and performance for PC gaming, professional graphics applications, and deep learning inferencing.

The Turing architecture has similar clock-speeds to the previous generation but includes an increase in the CUDA core count (average of 15% across the series), as well as a redesigned Streaming Multiprocessor (SM) architecture. These changes should enable greater shading efficiency, resulting in up to a 50% performance improvement per CUDA Core.

However, the real headline of the Turing architecture is the inclusion of RT and Tensor Cores.

RT Cores are accelerator units that are dedicated to performing ray tracing operations. They can be leveraged by systems and interfaces such as NVIDIA’s RTX ray tracing technology, and APIs such as Microsoft DXR, NVIDIA OptiX, and Vulkan ray tracing.
Tensor Cores power a suite of new deep learning-based neural services for games and professional graphics, in addition to providing fast AI inferencing for cloud-based systems. An example of this technology is the new Deep Learning Super Sampling (DLSS) capability, which aims to improve gaming image quality, whilst reducing the performance impact when compared against existing techniques.

NVIDIA believes the inclusion of RT and Tensor Cores will enable hybrid rendering, where ray tracing is combined with traditional rasterisation to exploit the strengths of both technologies.

The rest of this article will provide more detail regarding the potential impact of the RT and Tensor Cores, as well as my initial thoughts of the PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition graphics card.

If you are interested in the technical details of the Turing architecture, I would recommend reviewing the following two sources:

RT Cores

Real-time computer graphics have long used a technique known as rasterisation, where objects on the screen are created from a mesh of virtual triangles, or polygons, that create 3D models of objects.

Over the years, developers have become very efficient at rasterisation, achieving some incredible results as demonstrated in modern Triple-A games such as Red Dead Redemption 2, The Witcher 3, Far Cry 5, etc.

Unfortunately, rasterisation does have limitations. For example, due to the way rasterisation draws a scene, it can not correctly model the behaviour of a mirror or glass. This is because it only draws objects, it does not track the light itself.

In 1969, Arthur Appel (IBM) wrote the whitepaper “Some Techniques for Shading Machine Renderings of Solids”, which described a technique to use ray tracing to synthesise images in computer graphics.

Ray tracing is a rendering technique that can produce incredibly realistic lighting effects. In short, an algorithm can trace the path of light and then simulate the way that the light interacts with virtual objects in a computer-generated world.

This level of accuracy is considered by many as the “holy grail” of computer-generated images, producing results that are nearly indistinguishable from the real world.

Unfortunately, ray tracing is computationally expensive and (up until now) has been considered unfeasible for consumer hardware. The Turing architecture aims to change this, which (if successful) would have a profound impact on game development.

The video below is an example of ray tracing being used for reflections, powered by the Unreal Engine 4.

NVIDIA has a number of additional ray tracing video demonstrations available on their website.

Tensor Cores

As you might be aware, Tensor Cores are a key technology to a number of NVIDIA initiatives, specifically the Volta architecture used for Artificial Intelligence (AI).

The Turing architecture carries over the Tensor Cores from Volta, helping to speed up ray tracing by reducing the number of rays required in a scene by using AI denoising (a strength area of Tensor Cores).

Tensor Cores are also used to enable other capabilities, for example, Deep Learning Super Sampling (DLSS).

DLSS is a proprietary technology that uses deep learning to deliver better image quality at a lower cost than traditional Anti-Aliasing (AA) techniques, such as Temporal Anti-Aliasing (TAA).

The video below provides a good example of the DLSS vs. TAA.

Arguably TAA still produces a sharper image, however, the performance differences are certainly compelling (up to 33%).

PNY GeForce RTX 2080 Ti

Similar to the Geforce GTX 10 series launch, NVIDIA has released “founders editions” of the Geforce RTX 20 series. Unlike the Geforce GTX 10 series, each card includes a modest factory overclock. For example, the GeForce RTX 2080 Ti includes a boost clock of 1635MHz (up from the reference 1545MHz).

I have been testing the PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition graphics card, which includes the same factory overclock as the founders edition. The full specification can be found below:

PNY Part Number: VCG2080T11TFMPB-O
Card Dimensions: 1.50” × 11.88” × 4.50”
CUDA Cores: 4352
Clock Speed: 1350MHz
Boost Speed: 1635MHz
Memory Speed: 14Gbps
Memory Size: 11GB GDDR6
Memory Interface: 352-bit
Memory Bandwidth: 616GB/sec
I/O: 3x DisplayPort 1.4, 1x HDMI 2.0b, 1x USB Type-C
Power Input: 2x 8-Pin
TDP: 260W (Minimum 650W PSU)

The PNY card includes dual-slot cooling and is configured with two heatsinks and three fans. The cooling is 30cm in length, which is fairly large, extending 3cm past the PCB.

Common across all GeForce RTX cards, the PNY card includes an aluminium backplate. The card itself has a very subtle design, with no flashy extras such as RGB lighting (which I prefer).

The PNY card includes two 8-Pin power inputs, supporting the standard 260W TDP. However, expect the TDP to be higher when overclocking.

PNY decided to match the NVIDIA reference design for I/O, specifically 3x DisplayPort 1.4, 1x HDMI 2.0b, 1x USB Type-C. The USB Type-C can be used to simplify the use of VR Headsets.

Overall, the PNY card is very well built but lacks any unique selling point. PNY have chosen not to stray far from the NVIDIA reference design, even matching the modest overlock of the founders edition.

The two heatsink, three fan cooling design is definitely a benefit, which should result in greater overclocking potential. However, I suspect the results will be in line with other third-party cards that have chosen the same cooling configuration.

The RRP of the PNY card is £1299.99. At this price, I am not sure PNY have done enough to make their GeForce RTX 2080 Ti standout from the crowd. However, I have seen it on sale for £1199.99, which is more competitive when compared against other equivalent third-party cards.

Performance

I do not have a comprehensive test environment for graphics cards, therefore I would recommend referring to reviews from PC Perspective and Anandtech for a detailed analysis of the GeForce RTX 20 series.

With that said, I can confirm the GeForce RTX 2080 Ti is a beast! At 1440p and even 4K, it breezes through my modern game collection, generally above 100FPS at maximum image quality.

When added to my custom-built PC, the GeForce RTX 2080 Ti achieved a 3DMark score of 12137 (image below).

This is a healthy 71% performance improvement over the Gigabyte GeForce GTX 1080 it replaced. For reference, the GeForce GTX 1080 scored 7088.

Alongside the GeForce RTX 20 series release, NVIDIA also introduced the NVIDIA OC Scanner, which enables automatic overclocking and stress testing.

The NVIDIA OC Scanner takes approximately 20 minutes to complete and produces a result that NVIDIA state will be stable for daily usage. This is a welcome inclusion, allowing for “one-click” overclocking that will likely meet the needs of all but the most hardcore overclockers.

To access the NVIDIA OC Scanner, you will need a compatible overclocking utility. At the time of writing, only MSI Afterburner had been updated to include the new feature. Thankfully, overclocking utilities are not vendor specific, therefore I expect a lot of options to become available over the coming months.

Outlined below are the results of the NVIDIA OC Scanner, producing a new power curve and a new maximum reported boost clock of 2070MHz (up from the factory 1635MHz).

Unfortunately, it was not possible to test ray tracing or DLSS outside of controlled or synthetic benchmarks. As a result, I can not comment on the touted quality or performance benefits. In theory, some games (e.g. Battlefield V, Shadow of the Tomb Raider) will start gaining support for these features later in 2018 via software patches.

Conclusion

The GeForce RTX 20 series is an interesting release from NVIDIA. I commend them for continuing to push the boundaries of technology, especially considering the lack of competition from AMD. For example, it would have been very easy for NVIDIA to become complacent, essentially re-badging the GeForce GTX 10 series, knowing that their market position would remain secure.

The impact of hybrid rendering (ray tracing and rasterisation) is significant, however, developers will need to embrace and mature the techniques before we start to see tangible results. This is a concern, as the GeForce RTX 20 series will likely be reserved for a small percentage of the gaming population, recognising the high cost of entry and the relatively small market share of PC gaming, when compared against console gaming (e.g. PlayStation, Xbox, etc.)

As a result, the incentive for developers to prioritise techniques such as ray tracing feels like a stretch in 2018 and maybe 2019. Instead, they will likely focus on techniques that target the mass market and remain compatible with the current generation of consoles (e.g. PlayStation 4 Pro, etc.) This was evident from the launch of the GeForce RTX 20 series, as there was no game support for ray tracing or DLSS.

In conclusion, I believe the GeForce RTX 20 series has the potential to be a “game-changer” (literally). However, the high price and lack of software support will likely frustrate most consumers, especially considering the tangible benefits are likely multiple years away from maturity.

With this in mind, the GeForce GTX 10 series starts to look like a better deal (today), especially now that the cryptocurrency bubble has burst, resulting in more competitive pricing. The question is how long the stock will last, therefore, if you are in the market for a new graphics card, I would recommend moving quickly on a GeForce GTX 1070, 1080 or 1080 Ti.

Autonomous Vehicles

Mon, 24 Sep 2018 00:00:00 -0500

I recently completed a short residency at Stanford University as part of my executive education (see articles below).

As part of the residency, I took the opportunity to visit the Volkswagen Automotive Innovation Laboratory (VAIL), which is a research facility for students and researchers who are focused on developing vehicles of the future. Active projects include:

Research on drive-by-wire and driver assistance systems.
Research vehicles from the Stanford Artificial Intelligence Lab that use deep learning and computer vision to automate the driving process.
Research on the interaction of drivers with vehicles in the driving simulator.
Stanford Solar Car Project that designs, builds and races the latest vehicle competing in the World Solar Challenge.

Although the photo below looks like a standard Audi TTS, it is actually a fully autonomous vehicle, known as Shelley. The team shared how they have developed a neural network that allows Shelley to learn specific routes and adapt to different terrain (e.g. low friction, etc.)

As highlighted in the video below, the team tested Shelley at Thunderhill Raceway in the Sacramento Valley, where it successfully navigated the race track autonomously at speeds of 120mph.

To achieve this outcome, the Stanford team trained their neural network with data from 200,000 motion samples, controlled by high-performance graphics processors stored in the boot of the vehicle.

Recognising that 94% of crashes are caused by human error, autonomous cars have tremendous opportunity to improve safety, however, they must first be able to operate in unpredictable situations. The photo below highlights a student-built test bed, which is used to explore the limits of handling and grip in extreme scenarios.

The video below provides an overview of the test bed in use, navigating a track that is complicated by unexpected obstacles.

Finally, I took a quick photo of the VAIL driving simulator, which is similar to what is used by Formula One teams, but in this scenario focused on driver/vehicle interactions. I took it for a spin around a simulated city, which felt surprisingly accurate, considering the vehicle itself was stationary.

In conclusion, I remain incredibly excited about the future of autonomous vehicles and am eager to see who will be the first to go beyond the “driver assisted” capabilities (e.g. Tesla Autopilot) found in vehicles today.

CTO

Sun, 23 Sep 2018 00:00:00 -0500

In September 2018, I transitioned from my role as an Enterprise IT Architect to Chief Technology Officer (CTO).

As part of this major life event, I wanted to share my rationale and perspective on the new role.

Why CTO?

I have always been passionate about technology, both software, and hardware.

As a child, I would spend my time learning about every technology I could get my hands on and would dream about working for a company like Sony, helping to create the next great consumer product.

This ambition remained consistent until I reached high school, where I studied Business. This subject opened my eyes to other industries, where technology is not necessarily created, but applied to achieve specific outcomes. For example, taking the latest technology buzzword (e.g. Blockchain) and making it do something useful.

I quickly discovered that the “applied use of technology” was more exciting to me than the creation process, as it enabled me to explore many different technologies, whilst still in the pursuit of a specific outcome.

Alongside my passion for technology, I have always been goal orientated and highly competitive. I am attracted to complex, large-scale, high-pressure opportunities, even when I know it will likely lead to stress and anxiety. In short, being told something is impossible, only motivates me to work harder.

When I started my career (eleven years ago), I asked myself what role would best enable me to embrace my passion, providing me with the highest level of influence and accountability over key decisions.

It was at this point I first set my sights on the role of Chief Technology Officer (CTO).

What is a CTO?

The roles and responsibilities of a CTO are not consistently defined, covering everything from technology operations management to technology innovation (and sometimes both).

At the recent Gartner EA Summit, it was stated that 51% of today’s Fortune 500 CTOs are operationally focused, with the remaining 49% prioritising innovation. On average it takes twenty-four years for an IT Professional to become a CTO, covering eight different positions.

In my opinion, it is important to split operational and strategic roles, recognising that the required skill set and behaviours are very different. For example, I have observed phenomenal operational leaders, who have struggled to define and embed a coherent strategy (often impacting the long-term growth of the business).

Alongside the different behaviours, operational and strategic roles include intense, continuous responsibilities. If combined, the strategy is often neglected, as business-impacting operational issues must always take priority.

As a result, I believe the role of CTO is strategic, sitting at the intersection of business and technology. It requires a detailed understanding of the customer, technical breadth, and depth, as well as a proven record of enterprise delivery.

With this in mind, key responsibilities include:

Vision
Strategy
Enterprise Architecture
Innovation
Technology Investments
External Engagement

The role must cover the entire business value cycle (not limited to global services or infrastructure), partnering closely with the business function leaders.

Regardless of the industry, technology is now a core part of every business. This is evident from the “software is eating the world” revolution, where technology has been used as a catalyst to disrupt industries (e.g. Amazon, Uber, Airbnb, etc.)

As a result, I believe the role of CTO as a strategic leader is more critical now than ever before.

The Innovative Leader

Wed, 08 Aug 2018 00:00:00 -0500

This article builds upon the following two articles, summarising my experience at Stanford University.

In my previous article, I shared an overview of the Curriculum, Facilities, Professors, Participants, and Graduates. This article will focus on my key learnings.

NOTE: It would be very difficult to summarise the entire program, as a result, I have attempted to target specific areas that I believe offer tangible, real-world value.

Data vs. Intelligence

I often hear vendors and analysts state “data is the new currency”. However, data is only valuable when it enables insights and intelligence. By comparison, data is oil and insights are the petrol (gasoline).

As part of the session “Operational Excellence Through Flow Coordination”, Hau Lee discussed three flows that are key to operational excellence, specifically:

Re-engineering or redesigning these flows can lead to drastic improvements in operational excellence.

Hau Lee shared a number of interesting case studies, where companies have successfully leveraged or manipulated flows.

For example, 7-Eleven Japan is the number one seller of stockings in Japan.

Based on this information, the commercial team recommended that 7-Eleven expand their range into cosmetics, placing them next to the stockings. Without any additional information, this recommendation feels appropriate, as it is easy to assume that the primary customer of stockings would be female and therefore interested in cosmetics.

However, the CEO of 7-Eleven asked for additional information to be collected from the consumer, prior to any business decision. Specifically, he asked to understand the gender, approximate age and time of purchase. This additional information exposed that the primary customer of stockings was actually middle-aged men, purchasing at two specific times in the evening. The times were important, as it highlighted that these purchases were happening during peak commuting hours.

In summary, stockings in Japan are predominantly purchased by middle-aged men on behalf of their wives, during their commute home from work. Therefore, expanding into cosmetics would likely be a bad business decision.

I like this example, as it highlights how information reviewed in isolation, without context, is useless. Only when additional information is collected, can an intelligent business decision be achieved.

Premortem

One of the most powerful concepts I took away from the program was something called a premortem.

A premortem is a thought experiment for teams, which can be accomplished in a single meeting. At the point in which a key decision is about to be made, a premortem forces the team to look ahead one year and assume the deliverable has catastrophically failed. The team must then uncover what contributed to the failure.

The concept of a premortem was invented by Gary Klein and (as outlined in the video below) is backed by Economics Nobel Laureate Daniel Kahneman.

The process to complete a premortem can be summarised into five points.

When: A premortem must be completed just prior to a final decision being made. The decision could relate to a product, project, company, service, etc. The timing for the premortem is very important, as to early could result in incomplete information and demotivation, while too late may impact the ability to mitigate concerns.
Who: A premortem must be holistic, including representation from multiple teams and different levels across the organisation, helping to ensure a comprehensive analysis is completed. It also critical that all core team members are included in the exercise.
Why: A premortem must start by assuming that one year has passed since the deliverable and it has resulted in failure. Although the process sounds very negative, it is actually very cathartic, allowing the team to express their concerns, rather than simply ignoring them or relying on hope.
How: To ensure the analysis is balanced, every member of the group must start by independently capturing the reasons that contributed to the failure (similar to brainstorming). Once complete, the group can discuss and look to consolidate and refine the output, including key factors such as the likelihood and significance of the impact.
Next Steps: Once the group has a consolidated list of concerns, the discussion must switch to mitigation. Leveraging the Pareto principle (for many events, roughly 80% of the effects come from 20% of the causes), can help streamline the process.

In conclusion, a premortem is powerful as it tackles the risk of failure proactively and at multiple levels. For example:

It increases the likelihood of success by proactively identifying and mitigating concerns.
The process is cathartic, allowing everyone to vocalise their fears.
Promotes holistic transparency and alignment, covering multiple stakeholders and levels.
Improves teamwork by overcoming adversity, essentially giving the team an early (hypothetical) win.

Over the week, the professors shared a number of interesting premortem success stories.

For example, an airline completed a premortem on the company itself, looking twenty years ahead. The results highlighted that business travel (their primary revenue stream) would see a significant decline over the twenty-year timeframe. As a result, the Airline made the strategic decision to evolve their vision, focusing on “high fidelity meetings”, not specifically air travel.

Although subtle, this change is important, as it signifies to all stakeholders that the company will start to explore and invest in technologies outside of air travel, to support their long-term vision of delivering high fidelity meetings.

Design Thinking

As part of the program, we completed a bootcamp at the Hasso-Plattner Institute of Design (known as the d.school), which aims to build on methods from across the field of design to create learning experiences that help people unlock their creative potential.

The d.school is the leading university for teaching Design Thinking, which encourages organisations to focus on the people they are creating for and leads to human-centered products, services, and internal processes. A key resource is the Design Thinking Bootleg, which is a curated set of tools and methods to support Design Thinking.

As outlined in the image above, Design Thinking can be summarised into five stages.

Empathise: Gain an empathic understanding of the problem you are trying to solve by immersing yourself in the environment.
Define: Analyse your observations and synthesise them in order to define the core problems that you have identified. You should seek to define the problem as a problem statement in a human-centred manner.
Ideate: Start to “think outside the box” to identify new solutions to the problem statement you have created. Proactively look for alternative ways of viewing the problem.
Prototype: Produce a number of inexpensive, scaled down versions of the product or specific features. Investigate the problem solutions generated in the previous stage.
Test: Rigorously test the complete product using the best solutions identified during the prototyping phase.

It is important to note that the five stages are not always sequential, they do not have to follow any specific order and they can often occur in parallel and be repeated iteratively.

During the bootcamp, we practiced the techniques outlined in Design Thinking, following a prescribed scope focused on a wallet. The image below is an example of my work during the “Prototype” phase.

Finally, to demonstrate the potential impact of Design Thinking, the video below highlights how GE Healthcare embraced empathy to better understand how children experience an MRI scan.

In short, the negative, scary process of receiving an MRI scan (that often required sedation), was transformed into a positive, magical experience that promotes childlike wonder.

Emotional Brain

In the session “Decision Making: A Neuroscience Perspective” led by Baba Shiv, the group discussed how the rational brain is a slave to the emotional brain.

This is important to understand because when decisions are in inherently driven by emotion, if the emotion starts positive, the brain will rationalise the outcome regardless of the data points. With this in mind, it is possible to either manipulate outcomes or protect against emotional bias.

As outlined in the image above, the “X Framework” describes two types:

Type 1: Serotonin (Comforted) / Cortisol (Stressed)
Type 2: Low Dopamine (Bored) / High Dopamine (Excited)

Type 1 individuals are often the late adopters, very happy with the status quo, legacy systems, processes. Type 2 individuals are the early adopters, open to greater risk. It is also possible to be a hybrid, which includes individuals who are inherently excited about taking chances but need a nudge of reassurance.

It is important to note that context, not personality, determines the mindset (e.g. Type). For example, only when an individual has a stable level of comfort (Type 1), can they reach excitement (Type 2).

Apple understands the importance of the emotional brain. For example, at Apple Keynotes, they lead with “comfortable” information and customer stories, highlighting the success and stability of the company. This triggers comfort amongst the audience, which can then lead to excitement as they make new announcements. To ensure they also capture the “hybrid” individuals, they often close with reassurance, for example, a testimonial from a well respected individual such as Walt Mossberg.

In contrast, if Apple were to lead with the price of their products, they would likely trigger anxiety, resulting in a negative outcome.

Understanding the neuroscience behind decisions can enable better outcomes, recognising that at the individual or company-level, comfort is an important starting point.

Common Knowledge Effect

It is widely believed that bringing diverse people together will result in diverse (better) outcomes. In reality, laboratory studies have shown that information known to everyone prior to a discussion has a more powerful influence on decisions (compared to information that is not shared by everyone). This is known as the common knowledge effect and builds upon the human desire to “fit in”.

As highlighted in the image above, the diverse and/or individual thoughts are often lost or not openly represented.

To help combat the common knowledge effect, the following techniques should be considered.

Suspend initial judgment, do not complete open straw polls, etc.
Have everyone individually fill out a balance sheet (known facts and issues) prior to the meeting.
Focus on the problem to be solved, not a judgment.
Include a challenge team, someone who can play devil’s advocate.

Techniques that have been proven not to help include larger teams and/or more time.

The common knowledge effect is one of the reasons why Jeff Bezos (Amazon CEO) requires employees to individually create a story that can be reviewed prior to a meeting (replacing traditional slides).

Personally, I can recall multiple occasions where the common knowledge effect has likely impacted the outcome of a decision. This is certainly something I will be monitoring in the future and looking to combat with the techniques outlined above.

Conclusion

Overall, The Innovative Leader program at Stanford University was incredibly thought-provoking. The quality of the content, professors, facilities and peer group was world class and I relished the opportunity to learn and collaborate in this environment.

My goal now is to take this learning and put it into practice, starting with the areas I have outlined in this article.

MacBook Pro (2018)

Tue, 07 Aug 2018 00:00:00 -0500

Update - At the 30-OCT media event, Apple quietly announced an updated MacBook Pro with AMD Radeon Pro Vega graphics. This is a positive update (mitigating the concern I highlighted in this article), however, it was a little disappointing to see this update just three-months after the initial release of the 2018 model.

This article is a summary of my initial impressions using the 2018 15-inch MacBook Pro. The article is not intended to be a review, for that please refer to The Verge.

As previously highlighted, this device is my daily driver, supporting a diverse workload.

Productivity (O365, Atom, Sketch, OmniGraffle, ArchiMate)
Web Development (Node.js, Docker, Kubernetes)
Mobile Development (PWA and Xcode)
Game Development (GameMaker Studio 2)
Administration and Testing (VMware Fusion)
Photo Editing (Pixelmator)
Video Editing (Final Cut Pro)
Video Effects (After Effects)

Recognising the demands of software development (Docker, Kubernetes), as well as the need to run multiple virtual machines, I always target a high-performance processor, dedicated graphics and the maximum amount of supported memory. The full specification of my device can be found below:

15-inch MacBook Pro with Retina Display (2018)
2.9GHz 6-Core Intel i9 (Turbo up to 4.8GHz)
32GB 2400MHz DDR4 RAM
1TB PCIe SSD
Intel UHD Graphics 630 and AMD Radeon Pro 560X 4GB GDDR5
15.4-inch IPS Display (2880x1800, 220PPI)

I have split this article into four parts: hardware, performance, thermals and a conclusion.

Hardware

On the surface, the 2018 15-inch MacBook Pro is largely unchanged from previous generations (2016/2017), with four Thunderbolt 3 ports and a 3.5mm headphone jack.

The controversial butterfly mechanism keyboard has been upgraded to version three, which includes a new membrane (image below) to help prevent debris from impacting the long-term usability.

I have personally experienced multiple issues with previous versions of the butterfly mechanism, including unresponsive keys and double typing. As a result, I am hopeful that the new membrane will improve the reliability.

Regarding the actual typing experience, I would state that it feels very similar to previous versions, with a slight reduction in the noise. Although many people dislike the butterfly mechanism due to the short key travel, I have never found it to be an issue.

Alongside the keyboard, Apple also updated the Thunderbolt 3 controller, which now uses the new Intel JHL7540 chip, known as Titan Ridge.

This brings support for High-Bit-Rate 3 (HBR3), which is the signal standard for both DisplayPort 1.3 and DisplayPort 1.4. Therefore, the Radeon Pro 560X can drive up to two 5K displays at 60Hz (single stream) and theoretically supports an 8K display over a single cable.

Finally, Apple added the T2 chip, which was first introduced with the iMac Pro. The T2 chip consolidates a number of system components, specifically the System Management Controller (SMC), Touch ID, Image Signal Processor, Ambient Light Sensor, Audio Controller and the Mass Storage Controller.

The inclusion of T2 chip brings performance and security benefits, specifically a secure enclave that provides the foundation for encrypted storage (AES) and secure boot capabilities. It also adds native support for “Hey, Siri”, similar to what is found on the iPhone.

Performance

Unsurprisingly, performance is excellent. The upgrade from a 4-Core to 6-Core processor provides a healthy jump in performance for any software that can utilise multi-threading. This is validated by the impressive Geekbench scores.

Regarding memory, Apple has finally upgraded to DDR4, enabling a custom configuration up to 32GB. Knowing that DDR4 is more power hungry, Apple has also increased the battery capacity to ensure they still achieve a ten-hour estimate. In daily usage, I rarely utilise more than 16GB, however, I am pleased to see a 32GB option, as I occasionally work with multiple virtual machines for lab testing, etc. As an early observation, I have not seen a significant impact on the battery life, therefore I believe Apple has successfully offset the impact of the higher performance components.

Finally, it is worth highlighting the storage performance, which continues to be market leading. This is often an overlooked area, with devices like the Microsoft Surface Book 2 delivering dismal performance (600MB/s write). As a general rule, the overall performance of a computer is only as good as its slowest component (CPU, Memory, Storage, etc.)

As highlighted by the Blackmagic Disk Speed Test, the MacBook Pro storage performance is incredible, hitting 2700MB/s read and write (an increase of 500MB/s over the previous generation). I suspect this improvement is achieved through the T2 chip, which includes a new integrated Mass Storage Controller.

Overall, the 2018 15-inch MacBook Pro provides a healthy performance increase over previous generations. The one area that is a little disappointing is the dedicated graphics, which has not seen a significant upgrade since 2016. They have managed to deliver a minor performance increase through an improved 14nm FinFET process, however, the underlying architecture (Polaris) remains unchanged.

Thermals

Upon release, there was some controversy regarding the thermal performance of the Intel i9 processor. In short, it was unable to sustain base clock speeds (2.9GHz) under certain workloads. This issue was originally discovered by the Dave Lee (DAVE2D), who posted the following video:

Thankfully, Apple responded quickly and pushed a software patch to resolve the issue. Dave Lee posted a follow-up video (below), highlighting the improvements.

Thankfully, I received my MacBook Pro post-patch, therefore I did not personally experience the issue. However, it is worth recognising that Apple has always prioritised “thin and light” over raw performance. Therefore, high-performance components such as the Intel i9 processor will likely never operate to their maximum potential. Many people find this frustrating and even describe it as a design flaw. In my opinion, this is not a design flaw, it is a design decision.

As a result, if you plan to buy Apple hardware (specifically the MacBook range) you must accept that design will probably always be prioritised over raw performance.

Conclusion

Overall, I have been impressed with the 2018 15-inch MacBook Pro. The hardware feels more refined (third generation keyboard) and the inclusion of a 6-core processor, as well as a 32GB option, helps support the “Pro” designation.

Arguably, this is the device I had hoped would be released in 2016, when Apple first introduced the MacBook Pro with Touch Bar.

Finally, it is worth mentioning that the MacBook Pro remains a very expensive notebook. Therefore, I would argue that the “low end” MacBook Pro (2.2GHz i7, 16GB, Radeon 555X), with a 512GB Storage upgrade, would be the sweet spot for most people.

Stanford GSB

Mon, 06 Aug 2018 00:00:00 -0500

Last week, I completed a short residency at Stanford University as part of my executive education.

As highlighted in my previous article, I have been participating in the The Innovative Leader program at Stanford University Graduate School of Business.

The program targets executives who are in full-time employment, therefore, includes an intensive week on campus to connect, learn and collaborate.

Over two articles, I will attempt to summarise my experience, providing an insight into the program itself, the facilities and my learning. This first article will focus on the Curriculum, Facilities, Professors, Participants, and Graduates.

Curriculum

Outlined below is an example of the curriculum schedule.

As you can see, each day was fairly intense, with a combination of lectures, collaboration, assignments, and external engagement.

The program also included a comprehensive reading list (outlined below), with many of the books being written by the lecturers. Specific curriculum sessions build upon use cases and/or themes outlined in the reading list.

Some of these books are very well known, specifically “The Lean Startup”, “Thinking Fast and Slow”, etc. However, having personally helped to launch an emerging technologies group with my company, I found the book “Strategic Management of Technology and Innovation” very interesting, especially the examples outlined in the chapter “Managing Internal Corporate Venturing Cycles”.

Facilities

Stanford University is a private research university, currently positioned second in the world rankings. As a result, the facilities are incredible!

The campus itself is very large, as highlighted by the map below.

The majority of my time was spent at the Graduate School of Business, Knight Management Center (blue circle on the map).

The Knight Management Center is named after Philip Knight (co-founder of Nike), who earned his MBA at the school in 1962. It includes eight buildings and over one hundred classrooms, supporting a range of different learning styles (lecture theatres, flexible learning spaces, etc.)

The campus feels very “California”, embracing outdoor space, collaborative areas and art to inspire creative thinking. The business school’s slogan, “Change Lives. Change Organisations. Change the World.” is also prominently displayed throughout the campus.

I stayed in the Schwab Residential Center (red circle on the previous map), which is located across the street from the Knight Management Center and near the athletic facilities.

The Schwab Residential Center was rated first by the Financial Times for executive education food and accommodation. Personally, I would describe the accommodation as a cross between university halls and a hotel, meaning it was fairly basic. However, I can confirm that the staff and food were both excellent!

Due to the intensity of the week, I would highly recommend staying on campus, allowing for an immersive experience and providing the opportunity to engage in the extra-curricular activities, such as boot camps, campus tours, networking events, etc.

Professors

The Stanford faculty were world class, including accomplished business leaders, authors, and future Nobel laureates. The list of professors and their profiles can be found below.

Alongside the Stanford faculty, there were a number of guest speakers from start-ups, government agencies, venture capitalists and innovation incubators (e.g. Volkswagen Automotive Innovative Laboratory).

All of the Stanford faculty were excellent, however, I was specifically impressed by Baba Shiv (Program Director), Hayagreeva (Huggy) Rao, Hau Lee and Charles O’Reilly. Their expertise, experience and presentation style was very engaging, with thought-provoking topics and tangible outcomes.

Participants

In 2018, there were a total 64 executives completing the program, down-selected from 165 viable submissions.

The program was very diverse, with global representation (US, Germany, China, Australia) from enterprise companies (Google, Microsoft, Amazon), start-ups, government agencies and the military. For example, during lunch on the first day, I sat between a US Navy SEAL and a Lawyer, demonstrating that technology disruption can impact any industry.

As is often the case, one of the best parts of the program was the peer collaboration, allowing the group to compare stories, brainstorm ideas and discuss common barriers. My only regret would be the limited time available to engage with the peer community, which due to the intense schedule, was usually limited to lunch and dinner.

Graduates

During the week at Stanford, we also had the opportunity to meet with current Stanford graduates. Unsurprisingly, the quality of the research being completed was very impressive.

Outlined below are nine research topics that were presented to the group.

Ermelinda Porpiglia: High-resolution lineage mapping of myogenesis in vivo by single-cell mass cytometry.
HoJoon Lee: Analysis of cancer genomes for personalised immunotherapy in clinical phase 1 trail.
Sean Sketch: Simulating the impact of sensorimotor deficits on reaching performance.
Veronica Lin: Affordances and Challenges of Project Bloks.
Chuanxun Su: Construction of crystal structure prototype database.
Nasreddine El Dehaibi: Semantic classification for identifying sustainable content from Amazon product reviews.
Amr Shaltout: Spatiotemporal laser bean steering using frequency-arrayed optics in silicon metasurfaces.
Vivek Sunkari: Innovative formulations for prevention of chronic foot ulcers in diabetic patients.
Peng Si: Contrast-enhanced optical coherence tomography of tumour microvasculatures in vivo.

With my background in healthcare, I was specifically impressed with the work being completed by HoJoon Lee and Peng Si.

HoJoon Lee highlights the fact that certain somatic mutations in cancer genomes result in amino acid alterations. These mutations are unique to the individual patient, highlighting the need for personalised approaches to explore the interactions between T cell and mutation-derived tumour neoantigens. The research explored the advancements in immunotherapy and Next Generation Sequencing (NGS) technologies to enable precision cancer medicines.

Amr Shaltout is exploring the use of Optical Coherence Tomography Angiography (OCTA) to produce a three-dimensional image of human tissue by detecting the dynamic scattering produced by flowing red blood cells (RBCs). Due to the micrometer spatial resolution and millimetre tissue penetration depth, this approach could lead to a better understanding of tumour microenvironment, as well as improved monitoring of tumour progression and therapeutic response.

As you can probably tell from the topics, the ambition of the research happening at Stanford was very high. All of the research topics presented were very innovative, however, due to their specialist focus and complexity, I would challenge that additional time should be spent to ensure the premise and value proposition can be easily consumed (helping to secure future funding, etc.)

Conclusion

Hopefully, this first article has provided a good overview of Curriculum, Facilities, Professors, Participants, and Graduates. Overall, I was incredibly impressed with my Stanford University experience and feel fortunate to have had the opportunity to work alongside some of the worlds leading talent.

The next article will focus on my learnings.

Quantum Problems

Thu, 19 Jul 2018 00:00:00 -0500

Over the past couple of years, I’ve been investigating Quantum Computing. I have previously written two articles, covering an overview or the technology, hardware requirements, value proposition and the commercial model.

Similar to Blockchain, Quantum Computing is a complex technology that is largely misunderstood. As a result, I am always looking for new resources that tangibly explain its potential.

Earlier this week, Seeker published a new eight-minute video, which does a great job of explaining the technology and highlights the drug discovery opportunity.

One common misconception is that Quantum Computing will outperform Classical Computing at every task. In reality, Quantum Computing targets a specific set of tasks, which are categorized on the diagram below:

As you can see, scientists have identified a class of problem known as Bounded-error Quantum Polynomial (BQP), which can only be solved by Quantum Computing.

To better understand BQP, it is important to recognise the different complexity classes. Quantum Magazine recently posted a good article, which provides a high-level overview of the complexity classes. I have attempted to summarise the key points below.

Complexity Classes

A basic task of theoretical computer science is to sort problems into complexity classes. A complexity class contains all problems that can be solved within a given resource budget, where the resource is something like time or memory.

P = Polynomial

Short Description: All problems that are easy for a Classical Computer to solve.
Example: Typical problems include, identifying the shortest path between two points.

NP = Nondeterministic Polynomial

Short Description: All problems that can be quickly verified by a Classical Computer once a solution is given.
Example: The “Traveling Salesman Problem”. Given a list of cities and the distances between each pair of cities, what is the shortest possible route that visits each city and returns to the origin city?

PH = Polynomial Hierarchy

Short Description: A generalisation of NP. It contains all the problems you get if you start with a problem in NP and add additional layers of complexity.
Example: Determine if there exists a clique of size 50 but no clique of size 51.

PSPACE = Polynomial Space

Short Description: Does not place a focus on time, instead it contains all the problems that can be solved with a reasonable amount of memory.
Example: Every problem in P, NP, PH, and PS.

BQP = Bounded-error Quantum Polynomial Time

Short Description: All problems that are easy for a Quantum Computer to solve.
Example: Identify the prime factors of an integer.

EXP = Exponential

Short Description: All the problems that can be solved in an exponential amount of time by a Classical Computer.
Example: Contains all the previous classes (e.g. P, NP, PH, PSPACE, and BQP), however, problems categorised as EXP are not in P.

BPP = Bounded-error Probabilistic Polynomial Time

Short Description: Problems that can be quickly solved by algorithms that include an element of randomness.
Example: Two different formulas that each produce P, that have many variables. Do the formulas compute the exact same P? This is called the “Polynomial Identity Testing” problem.

Conclusion

Quantum Computing remains a fascinating area, which continues to challenge human understanding and therefore is in a constant state of evolution.

For example, BQP was first defined in 1993, encompassing all problems that can be solved by Quantum Computing.

Since then, computer scientists have hoped to contrast BQP with a class of problems known as PH, which encompasses all problems that can be solved by a Classical Computer. This required finding a problem that could be proven to be in BQP, but not in PH.

This breakthrough was achieved in May 2018, by computer scientists Ran Raz and Avishay Tal. Their findings were documented in the paper “Oracle Separation of BQP and PH”.

In my opinion, further breakthroughs in Quantum Mechanics are limited by the current level of software abstraction, which itself is limited by the hardware.

For example, future Quantum Computing hardware breakthroughs, such as Fault Tolerance and Coherence, will be followed quickly by greater software abstraction. This will result in scientists being able to identify and address increasingly complex problems.

This can be compared with the field of astronomy, which did not see major breakthroughs until after the creation of the telescope.

Babylon AI

Thu, 28 Jun 2018 00:00:00 -0500

UPDATE: Unsurprisingly, the statements made by Babylon Health (documented in this article) were considered highly controversial, trigging a passionate response from the medical community. As a result, the event video (embedded below) was made private. I have retained the embedded link, in the hope that it is republished.

Babylon Health aims to combine Artificial Intelligence (AI) with the medical expertise of humans to create a comprehensive, immediate and personalised health service that is universally available.

I recently shared my personal experience using Babylon Health, which enabled me to receive medical attention within hours, instead of days. This was achieved through the use of an AI-powered chatbot that triaged my symptoms, before automatically referring me to a GP via video messaging. The interaction was concluded through the creation of an online prescription that was made available for local pickup within the hour.

The experience was also a tangible example of the Healthcare Vision that I outlined in 2014, where technology is an embedded part of the healthcare ecosystem.

Yesterday, Babylon hosted an event at the Royal College of Physicians to reveal their latest advancements in AI. The full event (1h 30mins) can be viewed below.

At 35:35, Dr. Mobasher Butt shared how Babylon Health set the clinical standard for their AI, which incorporates the Membership of the Royal College of General Practitioners (MRCGP) curriculum and exam. They revealed that using a representative set of questions, their AI achieved a score of 81%, which can be compared to the average mark of 72% for a GP.

This is clearly an impressive feat, but was also met with scepticism, as the MRCGP exam questions are not publicly available. In response, Babylon Health highlighted that they used example questions, which are made available as a learning aid. They also encouraged the RCGP to share the full set of questions for future testing.

Babylon Health also touted the accuracy of their AI, using a set of vignettes (real-world scenarios). This included testing the AI against independent MRCGP qualified GPs. The results (outlined below) highlighted that the AI performed on par with the GPs included in the test, with an average of 80% accuracy.

Alongside the patient-facing results, Babylon Health shared a number of updates regarding their GP experience. This included the GP Portal, which is used during a patient interaction.

Not only does the GP Portal provide a comprehensive view of the patient’s history, but it also captures the conversation in real-time, looking to identify keywords to help support the diagnosis. This information can improve the speed and accuracy of the outcome, as well as significantly reduce the amount of time a GP must spend on documentation.

Another interesting demonstration was the use of Machine Learning (ML) to provide the GP with a real-time readout of a patient’s emotional state, which was achieved through analysis of the patents facial expressions. For example, a patient may express confusion or fear, but may not feel comfortable to share their feelings verbally.

As I reflect on the Healthcare Vision I outlined in 2014, I believe the advancements presented by Babylon Health demonstrate how this vision is rapidly becoming a reality.

For example, I predicted a world where a virtual assistant would be used to analyse health data, providing real-time recommendations. If a health issue was identified the virtual assistant would take one of the following actions:

If the condition is known, the virtual assistant would automatically order the required medication. This medication would be delivered via the Amazon logistics network.
If the condition is unknown or non-critical but requires consultancy, the virtual assistant would schedule a video meeting with a GP.

Babylon Health has clearly demonstrated that their virtual assistant (which can be integrated with Amazon Alexa) can accurately diagnose a set of symptoms. They also provide a fast and simple way to engage a GP via video messaging, with the potential to automatic prescribe medication as their AI gains credibility. When combined with the recent acquisition of PillPack by Amazon, the potential for rapid home delivery also becomes a reality.

In short, I am excited about the future of healthcare and see a tremendous opportunity for technology to enable better health outcomes, which are faster, more accurate and more accessible.

UK EA Forum - Part V

Sat, 26 May 2018 00:00:00 -0500

Yesterday, I facilitated the fifth UK EA Forum, which currently includes thirty-five people, across twenty-five companies.

The forum is open to any company in the UK and meets once per quarter to discuss common interests, opportunities and barriers. The details of past four meetings can be found below:

The agenda for each meeting is crowdsourced, ensuring the most relevant and topical areas are prioritised. Yesterday, the group discussed the following.

Enterprise Architecture
Multi-Cloud Positioning
Robotic Process Automation (RPA)
Microsoft Office 365 Positioning
General Data Protection Regulation (GDPR)

My notes from the meeting can be found below.

Enterprise Architecture

To launch the session, I shared our Enterprise Architecture structure, including principles and positioning.

Every company had a slightly different structure and level of maturity, however, there were many common themes, including the importance of placing architecture close to the business and ensuring strong sponsorship to enable strategic decision making.

The group also discussed the use of Enterprise Architecture tools to facilitate key processes. Some companies had no formal tool positioned, instead leveraging Visio and PowerPoint, while others had invested in a commercial offering. One company had custom-built a tool (Waltz, which is available on GitHub.

The specific Enterprise Architecture tools discussed can be found listed below:

The companies that were leveraging a tool described the importance of integration with service/portfolio management and highlighted the discipline required to maintain the data captured by the tool. As a result, an Enterprise Architecture tool should only be considered if the underlying processes are mature.

We also discussed the connection between Enterprise Architecture and Software Development. I shared our focus on standardisation (enforced by Continuous Integration) and accelerators (reusable code modules).

Finally, I demonstrated our dedicated Developer Portal, which is used educate the developer community (including service integrators, etc.)

Multi-Cloud Positioning

With the recent release of the 2018 Gartner Magic Quadrant for Cloud Infrastructure as Service (image below), the group discussed multi-cloud.

Source: ZDNet.

It should be noted that Gartner made a significant change to the Cloud IaaS criteria in 2018, with a new focus on global vendors that currently have hyper-scale integrated IaaS and PaaS offerings. As a result, the number of vendors went from fourteen in 2017 to just six in 2018.

The group discussed the merits of cloud abstraction, using open-source Infrastructure as Code capabilities such as Terraform to enable a multi-cloud strategy. On paper this approach makes a lot of sense, reducing lock-in and duplication, whilst promoting “utility compute”. However, the reality can be more complex, due to the speed in which cloud providers are innovating, resulting in API gaps.

Therefore, most companies had strategically positioned AWS and/or Azure for commodity services (e.g. Compute), with Google Cloud reserved for specialist capabilities (e.g. Cloud TPU).

Each cloud was supported using native tools (e.g. AWS CloudFormation), with the exception of container management, where a platform such as OpenShift or Cloud Foundry remains a popular choices due to the maturity over native Kubernetes (likely to change in the next three years).

Robotic Process Automation (RPA)

Robotic Process Automation (RPA) is another hot topic for enterprise customers, with the promise to drive significant cost savings through the automation of highly repeatable tasks.

The use of the word “Robotic” makes the technology sound very futuristic. However, RPA tools are actually very “traditional” and comparable to GUI testing tools such as Selenium, where a simple script is created to automate GUI interactions.

The video below highlights Automation Anywhere, arguably the most comprehensive RPA tool on the market.

The group discussed the importance of positioning, specifically when to use RPA vs Application/Business Process Modernisation.

The recent Forrester Podcast “RPA: A Building Block of Transformative Automation” provides a great summary of RPA and highlights a number of frameworks to help with positioning.

Microsoft Office 365 Positioning

Microsoft has made a lot of progress with Office 365, successfully migrating the majority of their existing enterprise customers.

Although the positioning of the core capabilities (e.g. Email, Calendaring, Productivity) is clear, Office 365 includes a lot of additional services, which often results in duplication. For example, Teams vs Slack, OneDrive vs Box, Planner vs Trello, etc.

As a result, each enterprise will need to define a process for positioning, recognising that Microsoft will continue to add new services to further enrich their ecosystem.

Support can also be a challenge, as traditionally, Microsoft services have been supported by “workplace” teams. However, many of these new services (e.g. PowerApps) require a different set of skills, which are often found in other teams.

The group did not have specific answers to these challenges but recognised the importance of maintaining proactive control, instead of letting Microsoft services virally grow without appropriate positioning (which is obviously part of the Microsoft growth strategy).

Coincidently, the GDPR enforcement date was also 25-MAY, therefore, the group spent a few minutes sharing updates regarding their status, covering people, process, and technology.

GDPR has been a long-running topic as part of the UK EA Forum and I expect approaches such as “Privacy by Design” to remain highly relevant over the coming months.

Conclusion

The next UK EA Forum is scheduled for August 2018. If you are interested in joining, please connect via our LinkedIn Group.

Surface Book 2

Thu, 03 May 2018 00:00:00 -0500

Over the past couple of months, I have been using the Surface Book 2 alongside my MacBook Pro. Although I have no intention of giving up my Mac, I do occasionally require a device that allows me to natively interact with the Windows eco-system.

With this in mind, I thought I would share a few observations, specifically focused on Performance and Build Quality.

Please note, this article is not designed to be a detailed review, for that, I would recommend PC Perspective, The Verge and/or TechRadar.

Performance

On paper, the Surface Book 2 is impressive, especially when compared to the MacBook Pro. The specification of my device can be found below, costing £2499.

13-inch Surface Book 2
1.90GHz Quad-Core Intel i7-8650U (Turbo up to 4.2GHz)
16GB 1866MHz LPDDR3 RAM
512GB PCIe SSD (Samsung PM961)
Intel UHD Graphics 620 and NVIDIA GTX 1050 2GB GDDR5
13.5-inch PixelSense Display (3000x2000, 267PPI)

Having access to a quad-core processor in a 13-inch device is great, especially when combined with dedicated graphics from NVIDIA.

This combination is made possible by the unique design of the Surface Book 2, where the CPU, Memory, and Storage are located in the display, leaving plenty of room for the dedicated graphics and additional battery in the base.

As a result, it is possible to detach the display and use it as a standalone tablet (sacrificing battery life and graphics performance). Although this 2-in-1 feature acts as a good differentiator, I have not found a practical use for the standalone tablet mode. It is simply too large and heavy when compared to a dedicated tablet, such as the Apple iPad.

Overall performance is good, especially from the CPU and GPU, which is validated by the strong PCMark 10 results.

However, like so many Windows devices, there are a few frustrating limitations.

For example, storage performance is disappointing. As highlighted below, the maximum write performance of the Surface Book 2 is 600MB/s, compared to 2200MB/s on my MacBook Pro.

I suspect this limitation is artificially imposed by Microsoft to protect battery life. The evidence being that performance has actually improved dramatically since launch through firmware updates, as a previous review from PC Perspective reported a write performance of just 300MB/s.

It is interesting that Microsoft has chosen to compromise in this area, recognising that storage performance is not generally disclosed as part of the marketing material, but does play a key role in overall device performance.

Another disappointing area is I/O, where Microsoft has decided not to include Thunderbolt 3. They have at least added USB-C, but this is limited to 5Gb/s and therefore can not be used to add an external GPU, etc.

You could argue that Thunderbolt 3 is not critical, however, on a device that costs £2499, I would expect to have modern I/O options, especially as the equivalent MacBook Pro includes four Thunderbolt 3 ports.

Build Quality

The build quality of the Surface Book 2 is excellent. In 2016, I made the following statement about the MacBook Pro.

“The chassis is both rigid and dense, whilst at the same time feeling incredibly refined. I honestly believe this is the best-manufactured notebook ever, delivering an unparalleled level of precision.”

Although not quite to the same extent, I believe the Surface Book 2 achieves a similar level of precision. One area where I would commend Microsoft is the keyboard, which is without question more durable than the MacBook Pro.

Other nice touches include the reassuring mechanical click that occurs when detaching the display and the subtle keyboard curve, which is designed to improve ergonomics. These small details provide a feeling of quality and I have seen no evidence (yet) that the hardware will degrade over time.

The controversial “hinge” is certainly unique and I would personally prefer the device to close fully, without leaving a gap.

However, I can not fault the functionality and it does give the device character, helping it stand out from the massive number of Windows laptops on the market.

My only criticism would be the top-heavy nature of the device, due to the majority of the hardware (CPU, Memory, Storage) being located in the display. This can result in a slight wobble when using the device on an uneven surface or when interacting with the touch-screen in laptop mode.

Conclusion

Overall, the Surface Book 2 is a good device, that could be great, if not for a few frustrating limitations.

Depending on your usage, you might be able to forgive the disappointing storage performance and lack of Thunderbolt 3, however, when combined with the £2499 price, it becomes a difficult device to recommend.

With that said, if you are looking for a true 2-in-1, the Surface Book 2 remains one of your only options and therefore you might be able to better justify the price premium. I, however, urge you to use the device as a 2-in-1 before purchasing, as I was unimpressed with the standalone tablet mode.

In conclusion, if you have any interest in the Surface Book 2, I would first explore the Lenovo ThinkPad X1 Carbon, Dell XPS 13, or even a MacBook Pro.

Fedora 28

Tue, 01 May 2018 00:00:00 -0500

I first started using Linux in 2001 (I was 16), driven by curiosity and the desire to further embrace the open-source community.

Over the years I have used many different Linux distributions, starting with Mandrake Linux 8.1.

Unfortunately, Mandrake Linux is no longer officially available, which is a shame, as it was an excellent choice for beginners, thanks to its incredibly comprehensive documentation.

Mandrake Linux 8.1 also supported a wide range of desktops, including KDE 2.2.1 and GNOME 1.4.1 (images below), which made experimentation easy and fun.

I initially favoured KDE (left image), but over the years transitioned to GNOME (right image).

Although I will always look back on Mandrake Linux with fond memories, I eventually switched to Ubuntu.

For many years, Ubuntu was everything I wanted from a Linux distribution, with strong sponsorship, frequent updates, and an active community.

Unfortunately, in April 2011, Ubuntu made the decision to introduce Unity as a replacement for GNOME. Although I believe Ubuntu had good intentions with Unity, I was not impressed by the early builds and personally felt the community would be better served by continuing to mature GNOME.

Thankfully, one of the best things about Linux is its flexibility. Therefore, instead of continuing to use Ubuntu, I simply switched (again) to Fedora, which (thankfully) had abandoned its early efforts to implement Unity.

Since switching to Fedora, I have been impressed with its speed and reliability, as well as the focus on cloud, development, security, and virtualisation (all things I value within my workflow). I have also learned to use Fedora “out of the box” with very little customisation, which makes rebuilding a machine very quick and easy.

With the recent release of Fedora 28, I thought I would share the process I follow to set up a new machine.

Fedora Setup

Outside of the pre-installed applications, I really only install a handful of third-party applications, specifically Chrome, Atom, Docker, Node.js, NixNote, 1Password, and Flatpak (which adds support for Dropbox, Slack, Spotify, and VLC).

Finally, I install Fedy and the GNOME Tweak Tool (AKA Tweaks), which both provide easy access to popular operating system themes, extensions, and advanced configuration.

The majority of the applications are installed using Dandified Yum (DNF), which is a software package manager for RPM-based Linux distributions.

Chrome

Although I have been impressed with the latest builds of Firefox, having Chrome available is still valuable (especially when using third-party extensions and Google services).

To install, run the following terminal command:

sudo dnf install fedora-workstation-repositories

Once complete, Chrome will be available to install from the Software store.

Atom

Atom is my preferred text editor. To install, simply head to the Atom website and download the latest RPM version.

Once complete, navigate the download directory and run the following terminal command:

sudo dnf install ./atom.x86_64.rpm

Once installed, Atom will be available for use.

Docker

Over the past few years, Docker has become an integral part of my development workflow. To install, run the following terminal commands:

sudo dnf -y install dnf-plugins-core

sudo dnf config-manager \
  --add-repo \
  https://download.docker.com/linux/fedora/docker-ce.repo

sudo dnf install docker-ce

You can check Docker is working by running the following two terminal commands:

sudo systemctl start docker
sudo docker run hello-world

If successful, you should receive the message “Hello from Docker!”

Node

Occasionally, I still have a need to run Node.js locally. To install, run the following command:

sudo dnf install nodejs

To confirm the installation, run node -v to display the version number.

NixNote

NixNote is a free, open-source note-taking application that natively synchronises with Evernote. To install, run the following terminal commands:

sudo dnf copr enable nunodias/nixnote2
sudo dnf update -y
sudo dnf install nixnote2

Once installed, click “Sync” to login to Evernote.

1Password

1Password is a popular and comprehensive password manager, available across all platforms (Windows, macOS, Linux, ChromeOS, iOS, Android, etc.)

The easiest way to install 1Password on Linux is via the Chrome Extension.

To install, simply download, run and login.

Flatpak

Flatpak is packaging technology, allowing for applications to be installed on any Linux distribution. The current list of compatible applications can be found on FlatHub.

Flatpak has been pre-installed since Fedora 24. Therefore, installation is as simple as downloading and running the Flatpak Repository File.

Once installed, compatible applications (Dropbox, Slack, Spotify, and VLC) can be installed directly from FlatHub.

Fedy

Fedy is a post-installation script written in bash specifically for Fedora. It provides simple access to additional applications, utilities, and codecs that do not come pre-installed with Fedora.

sudo dnf install https://dl.folkswithhats.org/fedora/$(rpm -E %fedora)/RPMS/fedy-release.rpm

sudo dnf install https://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm https://download1.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm

sudo dnf install fedy

Once installed, Fedy will be available for use.

GNOME Tweak Tool

The GNOME Tweak Tool can be installed from the Software store (simply called Tweaks).

Once complete, I would recommend installing the Shell integration for Chrome, which will make installing extensions very simple. To install, run the following terminal commands:

sudo dnf install chrome-gnome-shell

sudo dnf copr enable region51/chrome-gnome-shell

sudo dnf install chrome-gnome-shell

Open Chrome and browse to the GNOME Shell Extensions website. You will be prompted to install a Chrome extension.

Once complete, you can simply search and enable GNOME Shell Extensions directly from Chrome. I always install “Dash to Dock”, which includes advanced settings to customise the dock.

Cloudflare DNS

Sun, 01 Apr 2018 00:00:00 -0500

The Domain Name System (DNS) is the phone book of the Internet. It maintains a directory of domain names and translates them to Internet Protocol (IP) addresses. As a result, nearly everything that happens on the Internet starts with a DNS request.

By default, consumer DNS is delivered via an Internet Service Provider (ISP). For example, BT, Sky, Virgin, Plusnet, etc.

Unfortunately, these providers are usually slow, insecure and lack robust privacy controls, meaning anyone can see the sites being accessed, even if the content is encrypted. Worse still, some providers sell data about user Internet activity and/or use it to target ads.

An easy solution is to use a public DNS provider, for example Cisco OpenDNS (208.67.222.222 / 208.67.220.220) or Google Public DNS (8.8.8.8 / 8.8.4.4). These providers are generally faster, more secure and likely more trustworthy than an ISP.

I have been a long-time user of OpenDNS (pre-dating the Cisco acquisition in 2015), however it is worth acknowledging that both Cisco and Google are not completely unbiased, for example, 80% of Google’s revenue comes from advertising (although I’m not suggesting this impacts their public DNS).

Introducing Cloudflare DNS

Cloudflare runs one of the world’s largest, fastest networks. APNIC is a non-profit organisation managing IP address allocation for the Asia Pacific and Oceania regions.

Today, Cloudflare and APNIC have launched a new public DNS service, with a focus on speed, security, and privacy.

Performance

DNSPerf is an independent service for DNS performance analytics. As you can see from the image below, Cloudflare DNS is currently rated the fastest in the world, with a query speed of just 14.24ms.

The average ISP query speed is 65ms+.

Security

Even if you are visiting a site using HTTPS, your DNS query is sent over an unencrypted connection. This limitation also increases the risk of a “Man-In-The-Middle” attack, inadvertently directing traffic to a phishing, malware or surveillance site.

As a result, Cloudflare DNS supports HTTPS. Therefore, compatible operating systems, browsers, mobile applications, and network equipment (e.g. home router) can use the DNS over HTTPS endpoint instead of sending DNS requests in plaintext.

Privacy

Cloudflare DNS does not log user IP addresses (the way other providers do). As outlined below, their privacy policy is clear and very customer-centric.

Cloudflare will never sell your data or use it to target ads. Period.
All debug logs, which we keep just long enough to ensure no one is using the service to cause harm, of are purged within 24 hours.
Cloudflare will not retain any personal data / personally identifiable information, including information about the client IP and client port.
Cloudflare will retain only limited transaction data for legitimate operational and research purposes, but in no case will such transaction data be retained by Cloudflare for more than 24 hours.
Cloudflare will only retain or use what is being asked, not who is asking it.

Cloudflare has also partnered with KPMG to independently audit their systems annually to validate their privacy controls.

Setup Cloudflare DNS

Setting up CloudFlare DNS is easy. Simply, update the DNS configuration on your device (e.g. Laptop, Smartphone, Router) with the following IPv4 or IPv6 addresses.

IPv4:

1.1.1.1
1.0.0.1

IPv6:

2606:4700:4700:: 1111
2606:4700:4700:: 1001

Once complete, your DNS requests will be sent to Cloudflare DNS. To enable DNS over HTTPS, refer to the Cloudflare DNS documentation.

Conclusion

Cloudflare and APNIC state they were motivated to create a public DNS by their shared mission to “help build a better Internet”. This is certainly an admirable cause, although it is fair to state that Cloudflare could be accused of being biased (supporting their own network). As a result, it is great to see they are willing to be independently audited.

Although Cloudflare DNS was officially launched today, I did have some early access and can confirm it is working as designed. As a result, Cloudflare DNS is now my primary DNS provider.

More information about Cloudflare DNS can be found in the announcement blog post and by accessing 1.1.1.1 directly.

Ryze Tello Development

Fri, 23 Mar 2018 00:00:00 -0500

Earlier this month I purchased the Ryze Tello, which is low-cost drone (£99) that includes DJI Flight Technology and an Intel Image Processor for photos and video.

I have previously used the DJI Mavic Pro, which is a serious drone, at a serious price (£899). Although the Tello is obviously no match for the Mavic Pro, it definitely punches above its weight (and price).

The Tello is incredibly small, very well built and boasts an impressive specification. If you are familiar with DJI drones, the Tello looks like a smaller version of the DJI Spark (£449), with similar build quality.

The full specification of the Tello can be found below:

Weight: 80g
Dimensions: 98×92.5×41mm
Propellers: 3inches
Functions: Range Finder, Barometer, Vision System, Wi-Fi
Camera: 5MP (2592x1936), 720p 30FPS
Battery: 1.1Ah/3.8V
Flight Distance: 100m
Flight Speed: 8m/s
Flight Time: 13min
Flight Height: 10m

Considering the price, the Tello includes an impressive number of features, which are usually reserved for high-end drones:

iOS/Android App
Bluetooth Controller Compatibility
VR Headset Compatibility
Electronic Image Stabilisation Video
Auto Takeoff/Landing
Low Battery Protection
Failsafe Protection (Auto Land)
Collision Detection
Vision Positioning System (Stabilisation)
Auto Flight Modes (Throw and Go, 8D Flips)

The Tello is great fun to fly and thanks to the Vision Positioning System, is incredibly stable for a drone for its size (either inside or outside).

It also supports Scratch, which is an MIT-developed visual coding system that aims to help anyone learn the basics of programming.

Programming autonomous routines for the Tello is great fun, however, does require some setup.

Getting Started with Scratch Development

These instructions were written for macOS, but should also work for Windows.

Download and install Adobe Air.
Download and install the Scratch 2.0 Offline Editor.
Download an install node.js.
Download and unzip the “Tello.js” and “Tello.s2e” configuration files.
Open “Terminal” (or Command Prompt), navigate to the folder location that includes the “Tello.js” file and run the command node Tello.js.
Open Scratch and whilst holding the “Shift” key, click “File > Import Experimental HTTP Extension”. Select the previously downloaded “Tello.s2e” file.
All Tello development blocks will be found under “More Blocks”.
Connect to your Tello via Wifi, similar to what you would do with your iOS or Android device.

You are now ready to run your program, using an event (when space key pressed) programmed in Scratch.

The Healthcare Reality

Fri, 09 Mar 2018 00:00:00 -0600

In 2014, I wrote the article “The Healthcare Revolution”, which outlined my vision for the future of healthcare.

In this article (four years later), I will review the key technologies, looking to understand if they have become a consumer reality.

Vision vs Reality

“The floor measures your weight, body composition, including the percentage of bone mass, fat mass, muscle mass, and water mass, in addition to heart rate and pulse wave velocity (arterial stiffness).”

Although not yet embedded into your bathroom floor, the Nokia Body Cardio, which can be purchased for £149.95, is a consumer product that measures Weight, Fat Mass, Muscle Mass, Water Mass, Bone Mass and Heart Rate. It can also automatically detect multiple users.

Paired with the Nokia BPM+ and Nokia Sleep (coming soon), you have Blood Pressure and Sleep Quality (covering Heart Rate and Breathing).

This data can be easily complemented by more traditional health tracking mechanisms, such as a Fitness Tracker (e.g. FitBit) and/or SmartWatch (e.g. Apple Watch).

“You then use the toilet, providing instant analysis of your urine, which is screened for disease and infection.”

Home urine tests do exist, however, I haven’t seen anyone attempt to make this process “appealing” to consumers.

However, Thriva, a UK startup, have launched a consumer blood analysis service (costing between £24.00 and £69.00), which can provide a comprehensive picture of your overview health, including Cholesterol, Liver Function, Vitamin D, Iron Profile, B12, Diabetes, and Folate.

I have not personally used the service, but early reviews highlight that the process is quick and simple, whilst leveraging pre-existing, heavily validated and CE accredited partners and equipment (helping to ensure safety, quality, and privacy).

“Next, you shower, where a full body analysis is completed. For example, mole mapping via digital photo-dermoscopy surveillance.”

Unfortunately, I am not aware of any consumer-grade digital photo-dermoscopy surveillance services, which still require specialist equipment and expertise. However, you can only assume that as consumer cameras become increasingly sophisticated and widely available, that this type of service will become a possibility.

“You then brush your teeth, where your saliva is analysed to identify markers of endocrine, immunologic, inflammatory and infectious conditions.”

In 2017, a number of companies released smart toothbrushes, including established dental hygiene companies such as Oral-B and Colgate, as well as startups such as Playbrush. However, the primary focus is improved dental hygiene through more effective cleaning, not health screening.

With that said, when reviewed alongside consumer genetic testing services (such as 23andMe), which can identify genetic markers related to specific health conditions, it is easy to see how these technologies could be combined to provide a more holistic and continuous health service.

Today, genetic testing services are a one-time process, focused on DNA, however, the end-to-end procedure of collecting saliva and screening the results is certainly transferable, which, if automated as part of a daily activity could enable the type of outcomes highlighted in the vision.

“While brushing your teeth, your eyes are checked by your mirror, via retinal-scan analysis.”

Verily, a Google subsidiary, recently announced the ability to analyse the fundus (rear interior wall of the eye), enabling them to deduce an individual’s age, blood pressure, and whether or not they smoke. This data can then be used to predict health issues, such as the risk of suffering a cardiac event.

Although likely many years away from a consumer service, the technology is clearly maturing and could feasibly be integrated into a smart mirror.

“All of this data is securely and privately stored in the cloud, where your virtual assistant can analyse the results and provide recommendations.”

At CES 2018, virtual assistants such as Amazon Alexa and Google Assistant hit the prime time, being integrated into many different products. Apple has also entered the market with the HomePod, however, it currently lacks the mature, open eco-system of Amazon Alexa and Google Assistant.

I can easily see these virtual assistants becoming commonplace at home and work, acting as the central point for a wide range of services (including health). A key limitation will be security and privacy, which can hopefully be overcome through voice authentication and tight integration with other smart devices (smartphones, tablets, etc.)

“Automatically prescribe and order the required medication. If the condition is unknown or non-critical, but requires consultancy, your virtual assistant will schedule a telemedicine meeting with your defined health care professional.”

I recently experienced Babylon Health, a UK start-up focused on artificial intelligence, virtual consultations, and prescriptions. Their AI-powered chatbot already acts as an autonomous triage service and is being trialed by the NHS as an alternative to the “111” telephone helpline. However, their vision is to enable full diagnosis without any human intervention. As a result, Combining the AI-powered chatbot with a virtual assistant could be incredibly powerful.

Babylon Health can also already deliver prescriptions electronically, which, if integrated with Amazon Prime Now, could completely remove the need for the patient to leave their home, allowing them to focus on getting better.

“If the condition is known and requires immediate attention, an Uber will be requested with an “SOS” flag.”

Uber recently announced Uber Health, which will partner with healthcare organisations to provide reliable, comfortable transportation for patients.

This new service is almost exactly what I had predicted and is already available across 100+ healthcare organisations in the US.

Conclusion

Overall, many of the technologies I outlined in 2014 are now commonplace, some are delivered as consumer products (e.g. Nokia Body Cardio) and provide real-time output, while others are services (e.g. Thriva) that still require a level of human engagement, resulting in longer lead-times.

It is fair to say that none of the technologies have been tightly integrated into home life and very few successfully provide meaningful insights from the data they collect. However, I predict these areas will improve quickly as each technology becomes a commodity and new open ecosystems start to emerge.

Regulatory Change

Tue, 06 Mar 2018 00:00:00 -0600

I was recently a guest speaker as part of an industry webinar on regulatory change.

The webinar was triggered by the impending General Data Protection Regulation (GDPR) enforcement date (25-MAY-2018), which is considered to be the most significant change to data privacy in past twenty years.

The webinar itself did not specifically focus on GDPR, but used it as an example of why companies must evolve and become comfortable operating in a state of continuous change.

My section covered architecture, where I shared our methodology and technology foundation, which aims to provide us with increased flexibility, agility and inherently embed privacy, security, and quality.

The full webinar description is outlined below:

Regulatory change is one of the few constants in the pharmaceutical industry and with market forces pushing pharmaceutical companies “beyond the pill”, the stakes have never been higher.

Join Mavens for a webinar focused on how we are helping pharmaceutical organisations shift their perspective from viewing regulatory as a barrier to viewing regulatory as a potential competitive advantage.

Learn how we are helping clients like Eli Lilly rearchitect their business processes and information systems to meet the demands of the next GDPR; getting them to market faster with compliant digital solutions that build user trust and deliver amazing value.

The webinar is 55 minutes and my section starts at 29:30.

If you would like more information regarding the content that I shared, please refer to the blog posts “Enterprise IT Architecture”, “Principles”, “Positioning and Patterns”.

Babylon Health

Mon, 05 Mar 2018 00:00:00 -0600

The average wait time for a GP appointment in the UK is 13 days and unfortunately, the trend is only getting worse, with a 30% annual increase between 2016 and 2017.

Predicting illness is still unreliable, therefore when you are taken ill, you are already operating reactively, fighting to get better. In the majority of non-serious cases, you are advised to only speak to a GP if your symptoms do not improve after one week. At which point, a GP consultation is important to rule out anything more serious and (if required) be prescribed medication.

However, knowing that a GP appointment can take up to 13 days, you may actually end up suffering for several more weeks. This issue is compounded by the fact that the successful treatment of more serious conditions is often time sensitive.

Babylon Health

I was recently a little unwell, which resulted in me needing a GP appointment. I decided to make the appointment on Saturday evening, after feeling that my symptoms were not improving. However, my local GP only operates between 08:00 and 17:00 weekdays, therefore I would need to wait until the following Monday morning to make the phone call. At which point, I would join the GP appointment lottery.

However, I decided to try a different option…

Babylon Health is a health service provider (founded in 2013) that enables users to have virtual consultations with doctors and health care professionals via text and video messaging through its mobile app. It also supports drug prescriptions, referrals to health specialists, and health exams leveraging local facilities.

Babylon Health has also invested heavily in Artificial Intelligence (AI), starting with a chatbot that can act as an autonomous triage service, with the goal to eventually offer a full diagnosis.

The AI-powered chatbot is already gaining credibility, resulting in a number of health authorities looking to partner with Babylon Health. For example, the NHS is already exploring the technology as an alternative to the “111” telephone helpline, which patients call to get healthcare advice and out-of-hours medical services.

My Experience

NOTE: I have been researching Babylon Health for several years, but had never actually used the service. Therefore, I already had a verified account and the app downloaded on my iPhone.

At 19:06 on Saturday evening, I launched the Babylon app, where I was immediately greeted by the AI-powered chatbot. It politely asked me to complete nine questions, prior to making the suggestion that I book a GP appointment (validating my own suspicion).

This outcome automatically triggered the “book an appointment” form, which provides the option to select a time and (if required) add some notes, as well as images. At this point, the time was 19:12 and the next appointment was 19:20.

As soon as I booked the appointment, I received an onscreen confirmation (and e-mail), which included details about the GP, specifically her photo, location and medical background. The actual booking experience can be compared to booking an Uber, very intuitive.

At 19:20, my iPhone started to ring, asking me to join a video conference via the Babylon app. The GP introduced herself and had already reviewed my notes. The conversation lasted approximately 8 minutes and felt just like Apple’s FaceTime, with picture-in-picture video.

The conclusion of the appointment was a prescription for antibiotics. By 19:30, I had a message in my Babylon app, asking me to confirm my prescription details and provide a pick-up location. It used the iPhone GPS to locate my nearest pharmacy (Tesco) and informed me that the prescription would be ready within the hour.

At 20:30, I had taken my first dose of antibiotics, less than 1hour 30mins after deciding to book the GP appointment!

I was also able to review all of the GP notes, a full video recording of our conversation and the prescription (including dosage information) instantly from within the app.

Conclusion

1hour 30mins compared to anything from 3 to 13 days!

Overall, I have been incredibly impressed with Babylon Health. It successfully transformed a frustrating, old-fashioned and time-consuming process, into a seamless, modern and almost instant experience.

As a result, I would highly recommend Babylon Health to anyone. It is important to note that there is a cost (£50 per year for unlimited appointments), however, they are partnering with the NHS to trial free services for people living or working in London.

I truly believe I have experienced the future of GP interactions, which I suspect will only get better as the technology continues to mature.

My Setup (2018)

Fri, 16 Feb 2018 00:00:00 -0600

Update - Since writing this article I have made several hardware upgrades, specifically the MacBook Pro (including Radeon Pro Vega graphics) and my custom-built PC. As a result, this article has been updated to reflect the changes.

Final Update - Please refer to the article “My Setup (Q3 2025)” for a summary of my latest setup.

Home Setup

At home, I switch between three devices. My daily driver is a 15-inch MacBook Pro with Retina Display (OCT-2018). It’s primarily used for productivity tasks, software development, and photo editing. The full specification can be found below:

15-inch MacBook Pro with Retina Display (OCT-2018)
2.9GHz 6-Core Intel i9 (Turbo up to 4.8GHz)
32GB 2400MHz DDR4 RAM
1TB PCIe SSD
Intel UHD Graphics 630 and AMD Radeon Pro Vega 20 4GB HBM2
15.4-inch IPS Display (2880x1800, 220PPI)

I also have a 13-inch Surface Pro 2, which is a secondary device and primarily used for work, allowing me to natively interact with the Windows eco-system. The Surface Book 2 is a nice device, with great build quality, but limited by the surprisingly poor storage performance (300MB/s sequential write) and lack of Thunderbolt 3. The full specification can be found below:

13-inch Surface Book 2
1.90GHz Quad-Core Intel i7-8650U (Turbo up to 4.2GHz)
16GB 1866MHz LPDDR3 RAM
512GB PCIe SSD (Samsung PM961)
Intel UHD Graphics 620 and NVIDIA GTX 1050 2GB GDDR5
13.5-inch PixelSense Display (3000x2000, 267PPI)

Asus Maximus VIII Hero Alpha (Intel Z170)
Intel Core i7-6700K 4.0GHz @ 4.6GHz
Noctua NH-D15 CPU Cooler
Corsair Vengeance LPX 32GB DDR4 PC4-24000C15 (15-17-17-35)
PNY GeForce RTX 2080 Ti 11GB XLR8 Gaming Overclocked Edition
Samsung 950 Pro 512GB M.2 PCI-e NVMe SSD
EVGA SuperNova P2 1000W ‘80 Plus Platinum’ PSU
Phanteks Enthoo Evolv ATX Mid Tower Case
Bowers & Wilkins MM-1 Speakers

2x Dell U2718Q IPS HDR 4K (3840x2160 / 60Hz / 5ms / HDR10)
1x Dell S2716DG TN G-Sync (2560x1440 / 144Hz / 1ms)

My workspace is kept fairly minimal, with all monitors mounted on third-party stands to help maximise desk space.

As you can see, for input, I primarily use the Logitech Craft and Logitech MX Anywhere 2S. These peripherals were selected as they include “Easy-Switch”, which makes it simple to toggle between multiple Bluetooth devices.

The image below provides a closer look at the two Dell U2718Q IPS HDR 4K monitors and peripherals.

As highlighted in the image below, the PC is hidden under the desk, with the case door exposed providing easy access to the components.

The desk itself and matching pedestal is from Three Counties Office Furniture. They are both designed for corporate use, therefore are hard-wearing and include integrated cable management.

Finally, I use a Herman Miller Setu chair, which includes the Kinematic Spine to control resistance and automatically support your weight as you recline.

Work Setup

At work, I use the 15-inch MacBook Pro and 13-inch Surface Book 2 (depending on the workload). The MacBook Pro connects to two 27-inch LG UltraFine 5K Monitors, offering an impressive 5120x2880 resolution and P3 wide colour gamut.

The Surface Book 2 is generally used independently, as Windows support for high-DPI scaling and wide colour gamut remains frustratingly poor.

The desk is electric height adjustable and therefore can be used from a seated or standing position. Similar to home, it is kept very minimal.

I also use a Herman Miller chair at work, however, due to the long hours, it is the more robust Mirra 2. This chair includes a phenomenal amount of adjustment, ensuring all-day comfort.

The Linux Foundation

Thu, 25 Jan 2018 00:00:00 -0600

In recent months, my team and I have been working closely with the Linux Foundation, specifically the Hyperledger Project.

Our focus has been on a multi-party business process that tracks a list of reagents, solvents, and excipients in a GMP sequence from the starting materials to the drug product. The current process is inefficient, lacks transparency and is cost prohibitive, due to legacy license-based technologies. Therefore, we have been exploring Blockchain as a technology to help overcome these existing challenges.

Specifically, we have been testing our use case against technologies such as Hyperledger Fabric and Hyperledger Sawtooth.

We chose to work with the Linux Foundation to reinforce our commitment to open source, access their broad and influential ecosystem, as well leverage their expertise and experience.

Today, as a direct result of this work, we officially become a member of the Linux Foundation and the first pharmaceutical company to join the Hyperledger Project.

“Eli Lilly and Company is a global healthcare leader that discovers and develops medicines to help people live longer, healthier, more active lives,” said Mike Meadows, CTO, Eli Lilly & Company. “We believe healthcare is on the verge of a technology revolution. As a result, we are excited to join The Linux Foundation and Hyperledger, reinforcing our commitment to the open-source community and partnering on the potential for blockchain technologies within healthcare.”

Additional details regarding the membership can be found in the full Linux Foundation Press Release.

Due to the sensitivity of the use case, I am unable to share additional details at this time. However, we are keen to contribute to the open source community, therefore plan to share our technical experiences, feedback, and recommendations on Hyperledger over the coming months.

Jekyll PWA

Mon, 22 Jan 2018 00:00:00 -0600

I have previously written about Progressive Web Apps (PWA), which aim to combine the flexibility of the web, with the rich user experience of native apps.

LifeinTECH leverages the JAMstack web architecture, which promotes the use of JavaScript, APIs, and Markup. The goal is to remove the server-side runtime and CMS dependencies, shifting the focus to the client and services.

The specific architecture components of LifeinTECH are outlined below:

Docker (Automated Build)
GitHub (Source Code Management)
Jekyll (Static Site Generator)
jQuery (JS Library)
Sass (CSS Preprocessor)
Disqus (Comments)
Netlify (Web Host and CDN)
Let’s Encrypt (Certificate Authority)

In short, Docker is used to run the Jekyll build process, producing the static assets that are then pushed to GitHub and automatically deployed to Netlify.

To support my understanding of Progressive Web Apps, I set myself the challenge to upgrade LifeinTECH, aiming to achieve 100% compliance with the Google Lighthouse PWA audit.

Alongside the learning, I hope this upgrade will improve the end user experience of LifeinTECH, resulting in faster performance, offline access and native mobile features (push notifications, etc.)

I have documented the process I followed below.

NOTE: Prior to any development, it is important to note that Progressive Web Apps require SSL. This was a primary driver for my recent migration to Netlify, which simplifies the use of Let’s Encrypt.

Service Worker

As my build process is currently fully automated, I wanted to ensure I reduced the need for any manual steps. As a result, I found an excellent “Jekyll-PWA-Plugin” on GitHub, which installs as a RubyGem and automatically generates a Service Worker that provides pre-cache and runtime cache using Google Workbox.

The advantage of this plugin is that it has been designed to support the Jekyll workflow, therefore it can be easily configured to cache recent blog posts, etc.

To install the plugin, simply add jekyll-pwa-plugin to your Gemfile. You may also need to update your “_config.yml”, as outlined below:

gems:
  - jekyll-paginate
  - jekyll/tagging
  - jekyll-pwa-plugin

Once complete, the following configuration needs to be added to the “_config.yml” file:

pwa:
  sw_filename: sw.js
  dest_js_directory: js
  cache_name: pwa-cache
  precache_recent_posts_num: 12
  precache_glob_directory: /
  precache_glob_patterns:
    - "{js,css,fonts}/**/*.{js,css,woff,woff2}"
    - /index.html
    - /about.html
  precache_glob_ignores:
    - sw-register.js
  runtime_cache:
    - route: /^api\/getdata/
  strategy: networkFirst
    - route: "'/api/pic'"
  strategy: cacheFirst

This configuration prepares the Service Worker (sw.js), as well as copies the required “workbox-sw.prod.js” dependency into your JS directory.

Key configuration items include the number of posts to cache (in my case twelve), as well as the pre-cache glob pattern, where you can specify files, including images, etc.

Once configured, it is as simple as running the build process, which will automatically produce three files: “workbox-sw.prod.js”, “sw-register.js” and “sw.js”.

The “workbox-sw.prod.js” file will be located in your JS directory, while “sw-register.js” and “sw.js” will be located in your web root.

The produced “sw-register.js” file is outlined below. You can essentially ignore this file, as the code is largely generic and simply registers the Service Worker.

if ('serviceWorker' in navigator) {
    navigator.serviceWorker.register('/sw.js?v=1518012075').then(function(reg) {
        reg.onupdatefound = function() {
            var installingWorker = reg.installing;
            installingWorker.onstatechange = function() {
                switch (installingWorker.state) {
                    case 'installed':
                        if (navigator.serviceWorker.controller) {
                            var event = document.createEvent('Event');
                            event.initEvent('sw.update', true, true);
                            window.dispatchEvent(event);
                        }
                        break;
                }
            };
        };
    }).catch(function(e) {
        console.error('Error during service worker registration:', e);
    });
}

The “sw.js” file outlined below is more interesting, as it includes your custom configuration.

importScripts('/js/workbox-sw.prod.v2.1.1.js');

const workboxSW = new WorkboxSW({
cacheId: 'pwa-cache',
ignoreUrlParametersMatching: [/^utm_/],
skipWaiting: true,
clientsClaim: true
});
workboxSW.precache([{"url":"/js/jquery.fitvids.min.js","revision":"933ee1ba75e70b792d6d49d237cb620b"},{"url":"/js/jquery.ghostHunter.min.js","revision":"50ff499db8da8d530d813ff6f4f099ef"},{"url":"/js/script.min.js","revision":"f839794bc07ed5681695843c390e3f3f"},{"url":"/js/evil-icons.min.js","revision":"b77de0fe89ad34da6027b34963cb7c73"},{"url":"/js/jquery-2.1.4.min.js","revision":"f9c7afd05729f10f55b689f36bb20172"},{"url":"/fonts/Born/Born.woff","revision":"2d11011b3ebe66fbc7123e0f4e1a310a"},{"url":"/index.html","revision":"ccc8e26d2124603086b44a6fabfc00f6"},{"url":"/2018/01/25/The-Linux-Foundation/","revision":"aff5d2a51a3328e83d41379fa385851d"},{"url":"/2018/01/19/UK-EA-Forum-Part-Four/","revision":"05966de21a6763478073a83de9259105"},{"url":"/2018/01/14/Meltdown-and-Spectre/","revision":"ed58f70e47129c5339a965819d8eb1f9"}]);
workboxSW.router.registerRoute(/^api\/getdata/,
workboxSW.strategies.networkFirst());

workboxSW.router.registerRoute('/api/pic',
workboxSW.strategies.cacheFirst());

Here you can see the pre-cache, which for simplicity I have reduced to three blog posts. It includes the specific asset (article URL), as well as a hash value, which is used to ensure the cache remains current.

Finally, the Service Worker triggers a custom event (sw.update), when it has finished updating. This provides a simple mechanism to notify active users of any changes (via a toast notification, etc.)

This completes the Service Worker setup, however, you may need to spend some time tuning your cache, finding the right balance between user experience, performance, and storage.

Web App Manifest

The Web App Manifest is a simple JSON file (manifest.json), which is used to control how the app appears and how it is launched. The file itself is easy to create manually, however, you can also use a tool, such as the Web App Manifest Generator.

{
  "name": "LifeinTECH",
  "short_name": "LifeinTECH",
  "description": "Technology and Development",
  "lang": "en-GB",
  "start_url": "/",
  "display": "standalone",
  "orientation": "any",
  "theme_color": "#fff",
  "icons": [
    {
      "src": "images/icons/android-icon-36x36.png",
      "sizes": "36x36",
      "type": "image/png"
    },
    {
      "src": "images/icons/android-icon-48x48.png",
      "sizes": "48x48",
      "type": "image/png"
    },
    {
      "src": "images/icons/android-icon-72x72.png",
      "sizes": "72x72",
      "type": "image/png"
    },
    {
      "src": "images/icons/android-icon-96x96.png",
      "sizes": "96x96",
      "type": "image/png"
    },
    {
      "src": "images/icons/android-icon-144x144.png",
      "sizes": "144x144",
      "type": "image/png"
    },
    {
      "src": "images/icons/android-icon-192x192.png",
      "sizes": "192x192",
      "type": "image/png"
    },
    {
      "src": "images/icons/android-icon-512x512.png",
      "sizes": "512x512",
      "type": "image/png"
    }
  ],
  "background_color": "#fff"
}

Unfortunately, not every browser supports the Web App Manifest. As a result, some gaps need to be closed using traditional meta tags. Outlined below are the tags I included, as well as the link to the “manifest.json” file itself.

<link rel="manifest" href="/manifest.json">

<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
<meta name="mobile-web-app-capable" content="yes">
<meta name="apple-mobile-web-app-capable" content="yes">
<meta name="application-name" content="LifeinTECH">
<meta name="apple-mobile-web-app-title" content="LifeinTECH">
<meta name="theme-color" content="#fff">
<meta name="msapplication-navbutton-color" content="#fff">
<meta name="apple-mobile-web-app-status-bar-style" content="black-translucent">
<meta name="msapplication-starturl" content="/">

<link rel="apple-touch-icon" sizes="180x180" href="/images/icons/apple-icon-180x180.png">
<link rel="icon" type="image/png" sizes="512x512"  href="/images/icons/android-icon-512x512.png">
<link rel="icon" type="image/png" sizes="16x16" href="/images/icons/favicon-16x16.png">
<meta name="msapplication-TileColor" content="#fff">
<meta name="msapplication-TileImage" content="/images/icons/ms-icon-144x144.png">
<meta name="theme-color" content="#fff">

In the above example, I have not included multiple sizes for each icon. However, this might be advisable, depending on your target audience.

Conclusion

That’s it! Once you have created and configured a Service Worker and Web App Manifest, you should be well positioned to support the key features of a Progressive Web App. This can be confirmed by running the Google Lighthouse PWA audit from the Chrome DevTools.

As you can see from the results, LifeinTECH successfully passes the PWA audit, achieving 100% compliance. The Chrome DevTools also allow you to test and validate the individual improvements, such as performance, offline mode and push notifications, as well as interrogate the local storage.

In my opinion, the local storage (specifically pre-cache) was the most complex part of the process, attempting to find the right balance between experience, performance, and storage. In the end, I managed to achieve a pre-cache state of approximately 5MB, covering the “home” and “about” pages, as well as the most recent twelve posts (including all images).