Скачать или смотреть 6.03 - Job Schedulers

CHPC Student Cluster Competition 2022
Chapter 06: HPC Components 2
Instructor: Lara Timm

An HPC system might have thousands of nodes and thousands of users. How does the system decide who gets access to what resources and when? How do we ensure that a task is run with the specific resources that it needs? This job is handled by a special piece of software called a workload management system.

This chapter will provide you with an introduction to workload management systems - an essential component of HPC systems. We will cover topics including the fundamental concepts related to workload management, the roles of the job scheduler and resource manager, some common scheduling policies that may be used for job scheduling and an introduction into the commands and scripts used to manage jobs using common workload management tools - like PBS and Slurm.

By the end of this chapter you should have an understanding of workload management systems and why we use them. You should also have a basic idea to help you get started with submitting jobs on an HPC cluster!