| Abstract: |
|
A number of institutions have experience in setting up campus-scale clusters, but due to a variety of reasons, have seen a varying amount of success. System management resources are often limited, scheduling can be challenging, and platform upgrades are difficult to deploy . This BoF will review best practices for those setting up these internal HPC cloud deployments to share what has worked and not worked when configuring and managing these resources. The discussion will encompass storage environments, submit/compile environments, scheduler environments, software installation practices, and monitoring tools. Additionally, the discussion will capture what has worked well for people in building their clusters and what people wish they had done differently. |
|
