DescriptionModern research computing needs at academic institutions are evolving. While traditional HPC continues to satisfy most workflows, a new generation of researcher has emerged looking for sophisticated, self-service control of compute infrastructure in a cloud-like environment. Often, these demands are not for their own interest, but nonetheless present due to constraints imposed by data governance and protection policies that cannot be satisfied by traditional HPC.
To cater to these modern users, the Minnesota Supercomputing Institute deployed a cloud service for research computing called Stratus. Stratus is designed expressly to satisfy the requirements set forth by the NIH Genomic Data Sharing (GDS) Policy for dbGaP data. It is powered by the Newton version of OpenStack, and backed by Ceph storage. The service offers three features not available on traditional HPC systems: a) on-demand availability of compute resources; b) long-running jobs (i.e.,> 30 days); and c) container-based computing with Docker applications.