Rapid Re-Purposing: How the Foresight to Replicate Their HPC Cluster in the Cloud Fueled the University of Liverpool’s COVID-19 Research
SessionHPC in the Cloud
Event Type
State of the Practice Talk
Best Practices
Cloud and Distributed Computing
TP
TimeTuesday, 17 November 20201pm - 1:30pm EDT
LocationTrack 7
DescriptionThe Advanced Research Computing team at the University of Liverpool created a mission-critical cloud replica of their on-premises HPC cluster, named Cloud Barkla, to function as a rapid-deployment service in October, 2019. Originally deployed to assist the relocation of their on-premises cluster, the team had the foresight to build a cloud cluster that could be repurposed at speed, and the need for it came sooner than expected. Dr. Yalin Zheng and his team utilized this capability to launch their AI-based medical image analysis research for accurate diagnosis of COVID-19 on Cloud Barkla in record time to allow lead researchers Joshua Bridge and Yanda Meng to run scaling tests on their developed toolset in order to bring their concepts to peer review. This talk covers the rapid deployment of an HPC cluster fit for Dr. Zheng’s AI requirements, demonstrating how cloud offers flexibility within a complete HPC services offering.