Cloud Infrastructure solutions to run HPC workloads

Contributors:

• Sadaf Alam (Swiss National Supercomputing Centre)
• Tim Bell (CERN)
• Blair Bethwaite (New Zealand eScience Infrastructure)
• Martial Michel (Data Machines)
• Dhabaleswar Panda (Ohio State University)
• Timothy Randles (Los Alamos National Laboratory)
• Stig Telfer (StackHPC)
• Ildikó Vancsa (OpenStack Foundation)

---

Abstract:

Virtualisation and containers have grown to see more prominent use within the realm of HPC. Adoption of these tools has enabled IT organisations to reduce costs all the while making it easier to manage large pools of compute, storage and networking resources. However, performance overheads, networking integrations, and system complexity pose daunting architectural challenges. There is a growing trend for using OpenStack’s bare metal support to provide cloud-native APIs for flexible HPC infrastructure management. Meanwhile containers (another technology born of the cloud) are now being routinely deployed to Supercomputers to ease support of increasingly popular data analytics and machine learning workloads. OpenStack, Containers, and the orchestration thereof, all pose their own set of unique benefits and challenges.

This BoF is aimed at architects, administrators, software engineers, and scientists interested in designing and deploying cloud infrastructure solutions to run HPC workloads. Attendees can expect to learn first-hand the level and patterns of adoption and deployment of these technologies across the ISC audience, plus hear war stories and gain an understanding of where pain points continue. The panel represents a broad range of institutions with long experience in use cases in these areas. Ahead of the session the panel will consult with the community to create a set of interactive survey questions / polls, the results of which will then be made available and fed back into Open project teams.