The International Conference for High Performance Computing, Networking, Storage, and Analysis
The International Conference for High Performance Computing, Networking, Storage, and Analysis
Workshop:CANOPIE-HPC: Containers and New Orchestration Paradigms for Isolated Environments in HPC
Authors: Aldo Culquicondor (Google LLC)
Abstract: Kubernetes is a well established platform for easily and efficiently running serving workloads in clusters on prem and the cloud. With its increasing popularity, users have started porting batch and HPC workloads onto Kubernetes, encountering some challenges. In this talk, I will discuss the recent improvements we have done to the Job API to enable more batch and HPC workloads to run on Kubernetes and our collaboration with Kubeflow to enable MPI workloads at scale.