Birds of a Feather

This BoF (agenda link) is going to bring everyone up to speed with the current state of containers with an overview of the art-of-possible.

Key Information

The workshop takes place Monday 5/30 from 5:30AM to 6:30PM in hall E on the second floor.

Goals

Our goal is to provide an onramp for two types of HPC personas: 1. traditional HPC environments using a HPC scheduler (like SLURM) and a batch simulation workload (GROMACS, OpenFOAM, ...), 1. Research workflows (AI/ML, Bioinformatics, Data science) without a big system operations support who might be born in the cloud.

We'll show what those groups have in common (e.g. build an image that is able to run on as many sites as possible, pick the right binary for the job) and where they differ (tightly couple workloads using MPI, data management, shared environments).

Agenda

Segment Start Title Speaker Affiliation
Container Motivation 17:30 Container in High Performance Computing Shane Canon / Andrew Younge Lawrence Berkeley / Sandia National Labs
Focus: Run/Build 17:45 Container in SLURM environments and how to get started Jonathan Sparks HPE
High Performance Computing with Kubernetes Eduardo Arango RedHat
Where to get and how to build images Andrew Younge Sandia National Labs
Outlook / Q&A 18:10 Metahub: Contextual Container Registry Proxy Christian Kniep QNIB Solutions
Traps and Pitfalls to Avoid Everyone
Q&A
END 18:30 BOF closes