SimOps, A New HPC Market Category
Definition: “A market category is a specific segment of an industry where a product or service can be understood in relation to similar competitors and existing customer assumptions. It provides context for a product by grouping it with other products that share common characteristics or use, such as "CRM software" or "health products". By placing a company's offerings in a recognizable category, it helps customers quickly understand the business, and it helps companies define their target audience and differentiate themselves from competitors.”
At first glance, the current market of high performance computing (HPC) hardware, software, and services looks very diverse and multifaceted, not to say chaotic, especially for newcomers, sporadic users, and for outsiders. And it gets even more complicated when you want to build your own best-in-class HPC environment for your (usually highly complex) application simulation workflows, be it on premises or in the cloud. But there is a solution which we’ll describe in the following.
When we founded UberCloud in 2012 with the aim of building one B2B software platform running on any engineering application workflow and any HPC environment, we first started with the so-called UberCloud Experiments, so far 232 HPC Cloud projects with engineers and HPC experts around the world, collected in over 160 case studies published in almost 30 UberCloud Compendiums, example: The UberCloud Experiment - Technical Computing in the Cloud 2019. From the Lessons Learned and Recommendations collected in each of the case studies, we extracted 44 Best Practices (see 20 major examples here), mostly recipes on how to avoid HPC Cloud roadblocks and build the best suited software stack for a given application workload. The key to success was the use of custom HPC containers and Kubernetes.
In early 2024, we discovered that DevOps (in 2008) and FinOps (in 2018) faced similar challenges which they tackled with building a Framework consisting of a set of principles, tools, and practices that promote collaboration between development and operations teams to automate and streamline the software delivery lifecycle. Key components include people, processes, and products (or technology), which work together to improve efficiency, speed, and quality through practices like continuous integration and continuous delivery (CI/CD) and automation.
On the other hand, we found that the engineering simulation and the HPC infrastructure communities face the same problems which affected people, processes, and products (or technology) which often lack collaboration, efficiency, speed, and product quality. Especially when making wrong choices about software and services.
And that was the “hour of birth” for SimOps, Simulation Operations Automation. Based on 44 best practices extracted from our 232 UberCloud (now SIMR) Experiments, and DevOps as a poster child, we developed the shared unified SimOps Framework, shaped by the engineering community to improve how simulations are run, managed, and scaled. Built on collective experience and proven practices, SimOps empowers teams to collaborate more effectively, automate repetitive tasks, and unlock innovation through accessible, reliable simulation workflows. And every software which contributes to these SimOps benefits will become part of the “SimOps Software Stack” and the new SimOps market category (as a subset of the much wider general HPC market).
