DescriptionHigh Performance Computing continues to rise and is no longer the domain of the special centers. It is inevitable that it will place demands on facilities and provide for the care and feeding of the external resources needed. It is common for HPC admins to get tasked at the last minute, with planning the data center needs specific to HPC. This tutorial addresses that since there is no other course available to impart the basic HPC-centric facilities planning. We know how to run codes efficiently but not to how to manage the surrounding infrastructure planning process. Data center design is critical as otherwise HPC associated expensive resources can be hobbled by lack of proper data center facilities.
In this tutorial, the authors propose to train users on how to begin the planning process, how to take stock of the current environmental demands, space, cooling, electricity, security, operations, and maintenance as well as foresee potential problems, point out methods and sources helpful to diagnose shortcoming in existing or incomplete data centers and develop data and insight necessary to distinguish and document correct and incorrect implementations.