Metrics for parallel job scheduling and their convergence

Dror G. Feitelson*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

71 Scopus citations

Abstract

The arrival process of jobs submitted to a parallel system is bursty, leading to fluctuations in the load at many time scales. In particular, rare events of extreme load may occur. Such events lead to an increase in the standard deviation of performance metrics, and thus delay the convergence of simulations used to evaluate the scheduling. Different performance metrics have been proposed in an effort to reduce this variability, and indeed display different rates of convergence. However, there is no single metric that outperforms the others under all conditions. Rather, the convergence of different metrics depends on the system being studied.

Original languageAmerican English
Title of host publicationJob Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers
EditorsDror G. Feitelson, Larry Rudolph
PublisherSpringer Verlag
Pages188-205
Number of pages18
ISBN (Print)3540428178
DOIs
StatePublished - 2001
Event7th International Workshop on Job Scheduling Strategies for Parallel Processing, JSSPP 2001 - Cambridge, United States
Duration: 16 Jun 200116 Jun 2001

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2221
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference7th International Workshop on Job Scheduling Strategies for Parallel Processing, JSSPP 2001
Country/TerritoryUnited States
CityCambridge
Period16/06/0116/06/01

Bibliographical note

Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 2001.

Fingerprint

Dive into the research topics of 'Metrics for parallel job scheduling and their convergence'. Together they form a unique fingerprint.

Cite this