Building a Supercomputer in the Cloud with Google Kubernetes

PGS Chief Enterprise Architect, Louis Bailleul, recently presented at the Google Cloud Next 22 event held in Munich. A related YouTube interview was given on the PGS journey into cloud-based supercomputing with Maciek Ruacki, Senior Project Manager, GKE, Google Cloud. Louis discusses the best-practice execution of batch and high-performance computing workloads on Google Kubernetes Engine (GKE) and how PGS used these to replace their 202 000 core Cray supercomputers in Houston. When used at peak capability, the GKE supercomputer running in the Google Cloud Platform (GCP) has a hypothetical 72.02 petaFLOPS capability, equivalent to the 7th largest computer in global rankings. Elements are discussed of the move and improve journey PGS has undertaken from on-premise Cray supercomputers to the GKE supercomputer available today in the cloud.