Isolating Cluster Jobs for Performance and Predictability
I hope everyone enjoyed their holiday festivities this week. You might be running out the door this very moment to spend that gift cash/cheque. Stop! Check out our next speaker first. You just might be inclined to spend those greenbacks on a front row seat to hear Brooks Davis talk about his work with FreeBSD computing clusters…
At The Aerospace Corporation, we run a large FreeBSD based computing cluster to support engineering applications. These applications come in all shapes, sizes, and qualities of implementation. To support them and our diverse userbase we have been searching for ways to isolate jobs from one another in ways that are more effective than Unix time sharing and more fine grained than allocating whole nodes to jobs. In this paper we discuss the problem space and our efforts so far. These efforts include implementation of partial file systems vitalization and CPU isolation using CPU sets.
Brooks is an Engineering Specialist at the Aerospace Corporation and a hobbyist blacksmith. As such, I expect a riveting discussion while he forges through red-hot points of interest and hammers home the concepts of clustering performance and predictability. Steel interested? Register for DCBSDCon 2009 now!
