Major trends have brought the cloud and high performance computing (HPC) communities closer together: the maturation of cloud technologies has shifted focus toward running workloads efficiently, and composite scientific workflows and resource heterogeneity and dynamism have revealed the limitations of traditional HPC resource and workflow management. It has become clear that the initially disparate communities have much to benefit in working together. Communication is a big part of that. What we’ve realized in our work is that when discussion happens between communities, the same terms might be used but have different meanings. For example, a “service” in Kubernetes means something very specific, while in HPC it might be generally describing a database or application programming interface. The term “scheduler” and what that encompasses also varies quite a bit between communities.
This documentation site aims to bridge the gap between HPC and cloud to discuss terms for converged computing – the collaborative space between these traditionally separate communities – to develop novel technologies and applications. These hybrid technologies might span the gamut from automation, workflows, containerization, to software development and deployment and testing. This is a timely topic as collaborative work is happening to a greater degree that combines approaches from high performance computing with cloud-native computing. Being able to communicate effectively is essential.
You can read the project README to see the original authors of the terms that are not represented in the GitHub commits here.
Converged Computing Authors. "About." Converged Computing Community Space 29 Nov 2024, https://converged-computing.org/about (accessed 29 Nov 24). [doi]