Core Algorithms of the Maui Scheduler

Abstract

The Maui scheduler has received wide acceptance in the HPC community as a highly configurable and effective batch scheduler. It is currently in use on hundreds of SP, O2K, and Linux cluster systems throughout the world including a high percentage of the largest and most cutting edge research sites. While the algorithms used within Maui have proven themselves effective, nothing has been published to date documenting these algorithms nor the configurable aspects they support. This paper focuses on three areas of Maui scheduling, specifically, backfill, job prioritization, and fairshare. It briefly discusses the goals of each component, the issues and corresponding design decisions, and the algorithms enabling the Maui policies. It also covers the configurable aspects of each algorithm and the impact of various parameter selections.