OCF Steel Stack is a comprehensive software suite designed to streamline High Performance Computing (HPC) deployments.
OCF Steel Stack is a comprehensive software suite designed to streamline High Performance Computing (HPC) deployments. As we prepare for the upcoming OCF Steel Stack 26 release, we reflect on its origins, the present-day capabilities, and future direction
Let’s dive into each one a little deeper >>
STEEL STACK AUDIT
At its birth in 2021, OCF Steel Stack provided a toolkit for customers to utilise their High Performance Computing (HPC) resources.
The main components of OCF Steel Stack are:
>> Provisioning | Repositories | Configuration Management | Documentation
A brief summary of the installation process is this:
Alongside this, documentation was provided for:
Since joining OCF two years ago, I have gained a detailed understanding of the intricacies of OCF Steel Stack – effectively conducting an ongoing audit. There have been several positive outcomes from this, as well as some challenges en route. For example – migrating our documentation.
DOCUMENTATION
Steel Stack documentation is now hosted online, managed via Git source control and maintained by the OCF Research and Development team. It evolves alongside feature development and is accessible to customers through a web browser. The transition from static formats (Word/PDF) to a dynamic, Markdown-based system began in late 2023, improving update efficiency and collaboration across the engineering teams.
We use MKDocs with the ‘Material’ theme for documentation – benefits include being open source, widely used and providing a huge range of features.
CONFIGURATION MANAGEMENT
In 2023, OCF Steel Stack provided a choice of two configuration managers for customers to choose from – Salt (or SaltStack) and Ansible. This required OCF engineers to work with both; resulting in context-switching when undertaking administration or installation on different clusters. We quickly identified that ‘managing two configuration managers’ was going to cause problems. For the OCF Research and Development team, it meant labour-intensive duplication of bug fixes.
Salt has advantages, particularly that it runs quickly. Originally, it was the configuration manager of choice for OCF Steel Stack deployments. However, the industry was moving more and more towards Ansible – this was a major factor in Ansible being subsequently incorporated into Steel Stack alongside Salt.
There were other reasons for the adoption of Ansible:
RIGOROUS TESTING
The OCF Research and Development test all bug fixes and new features before releasing them into production, using several test environments. One challenge we are considering is how to ensure that all customers can effectively update to the latest version of OCF Steel Stack and still keep any site-specific changes made to their cluster.
We currently have an ‘upgrade mechanism’ to do this. Customers can upgrade their own cluster or utilise the highly-skilled and experienced OCF managed service engineers and compute engineers, who can assist in ensuring that bespoke site customisations are preserved.
Possibly the most exciting part of working with Steel Stack is thinking about future developments.
‘T’ MINUS 100 FOR OCF Steel Stack v26
The next major release for OCF Steel Stack is scheduled for Q1 2026.
Our last major release was in April 2025 (v2504.0, with various bug fixes bringing us to the current version 2504.3). OCF Steel Stack 26 is now around 100 days away! The release process will involve extensive end-to-end testing, as already mentioned, along with associated Documentation updates.
We intend to bring in some exciting new features as well as enhance the currently existing code.
OCF Research and Development, in conjunction with OCF engineers and the wider OCF team, have identified priority features for the upcoming release.
Planned changes include:
The long-term strategy for OCF Steel Stack is always being considered and reviewed. We are in the early stages of investigating how containerisation could benefit future deployments.
Infrastructure-as-code (IaC) planning is underway for rapid, repeatable, consistent deployment of HPC resources. We are watching, with keen interest, the development of IaC tools ‘Terraform’ and ‘OpenToFu’. The OCF Research and Development team will continue to compare the directions of these two since OpenToFu split from Terraform in 2023, and the features of each product as they develop.
Our ambition is to be the most secure HPC vendor in the industry. We are confident that our security measures are robust – but in the ever-changing technological landscape, we must ensure that security is continually being enhanced. We have recently made some significant improvements to the operational security of OCF Steel Stack and intend to continue at pace with this as we move forward.
… look out for more information as the release of OCF Steel Stack 26 approaches!