Teaching HPC Systems Administrators

Abstract

The ability to grow and teach systems professionals relies on having the capacity to let students interact with supercomputers at levels not given to normal users. In this paper we describe the teaching methods and hardware platforms used by Purdue Research Computing to train undergraduates for HPC systems-facing roles. From Raspberry Pi clusters to the LittleFe project, previous work has focused on providing miniature hardware platforms and developing curriculums for teaching. Recently, we have developed and employed a method using virtual machines to reach a wider audiences, created best practices, and removed barriers for approaching coursework. This paper outlines the system we have designed, expands on the benets and drawbacks over hardware systems, and discusses the failures and successes we have had teaching HPC System Administrators.

BEAST Lab: A Practical Course on Experimental Evaluation of Diverse Modern HPC Architectures and Accelerators

“…FreeCom-pilerCamp.org of Lawrence Livermore National Laboratory [21] is an open online platform designed to train researchers in developing OpenMP compilers primarily to address the lack of training resources for researchers who are involved in the compiler and language development around OpenMP. Another worth-mentioning work is on teaching methods and hardware platforms used by Purdue Research Computing for educating HPC system administrators [23]. Overall, we can conclude that providing resources to users for training is a central piece of the teaching method in most of the existing programs.…”

Let's Get Our Heads Out of the Clouds: A scalable and sustainable approach to HPC Training Labs for Resource Constrained Environments and anyone else stuck in the Clouds

“…Within the global HPC community, the common approach to delivering HPC System Administrator training is through physical face-to-face engagements. The hands-on component of HPC training is facilitated through leveraging existing HPC infrastructure or cloud-based services that simulate an HPC environment [2,[7][8][9][10]40].…”