ABSTRACT: When CERN (European Organization for Nuclear Research) started creating a new cluster for the Large Hadron Collider (LHC), it needed to quickly scale up computing power while keeping costs in check. The result? The world’s largest and highest-energy particle accelerator is now supported the world’s largest HPC cloud. This BOF discussion will detail how CERN, in partnership with Platform Computing, built an open, cost-effective platform that would allow the organization’s global network of scientists and researchers to manage both virtual and physical servers in the cloud. This initiative has enabled CERN to improve CPU utilization, empowered scientists to manage their own processing workloads, and greatly reduced the amount of resources and servers required to power the LHC cluster. Discussion participants with similar challenges involving large datasets, massive throughput requirements and finite budgets will learn how to leverage the cloud to squeeze more capacity from existing infrastructures and hear best practices for evolving their current HPC environments. They will also be invited to share their own challenges, successes and lessons learned.
Session Leader Details:
William Lu (Primary Session Leader) - Platform Computing