A Methodology to Scale Containerized HPC Infrastructures in the Cloud

This paper introduces a generic method to scale HPC clusters on top of the Kubernetes cloud orchestrator. Users define their targeted infrastructure with usual syntax for recipes, and our approach automatically translates description full-fledged containerized cluster. Moreover, resource extensions or shrinks are handled, allowing dynamic resize cluster without disturbing its running. The orchestrator acts as provisioner. We applied three orthogonal architectural designs Open Source schedulers: SLURM, OAR, OpenPBS. Through series experiments, demonstrates potential regarding scalability issues simultaneous deployment several job schedulers in same physical infrastructure. It should be noticed that plan does not require any modification either containers schedulers. Our proposal is step forward reconciling two ecosystems cloud. also calls new research directions concrete implementations consolidation servers sober placement policies at level. works contribute running environment test technique robustness by adding removing nodes fly.