Metrics for parallel job scheduling and their convergence

Dror G. Feitelson

doi:10.1007/3-540-45540-x_11

Metrics for parallel job scheduling and their convergence

Dror G. Feitelson^*

^*Corresponding author for this work

The Rachel and Selim Benin School of Engineering and Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

72 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 language	English
Title of host publication	Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers
Editors	Dror G. Feitelson, Larry Rudolph
Publisher	Springer Verlag
Pages	188-205
Number of pages	18
ISBN (Print)	3540428178
DOIs	https://doi.org/10.1007/3-540-45540-x_11
State	Published - 2001
Event	7th International Workshop on Job Scheduling Strategies for Parallel Processing, JSSPP 2001 - Cambridge, United States Duration: 16 Jun 2001 → 16 Jun 2001

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	2221
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	7th International Workshop on Job Scheduling Strategies for Parallel Processing, JSSPP 2001
Country/Territory	United States
City	Cambridge
Period	16/06/01 → 16/06/01

Bibliographical note

Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 2001.

Access to Document

10.1007/3-540-45540-x_11

Cite this

Feitelson, D. G. (2001). Metrics for parallel job scheduling and their convergence. In D. G. Feitelson, & L. Rudolph (Eds.), Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers (pp. 188-205). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2221). Springer Verlag. https://doi.org/10.1007/3-540-45540-x_11

Feitelson, Dror G. / Metrics for parallel job scheduling and their convergence. Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers. editor / Dror G. Feitelson ; Larry Rudolph. Springer Verlag, 2001. pp. 188-205 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{3e1decb66a0d46798100da04b19e521a,

title = "Metrics for parallel job scheduling and their convergence",

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.",

author = "Feitelson, {Dror G.}",

note = "Publisher Copyright: {\textcopyright} Springer-Verlag Berlin Heidelberg 2001.; 7th International Workshop on Job Scheduling Strategies for Parallel Processing, JSSPP 2001 ; Conference date: 16-06-2001 Through 16-06-2001",

year = "2001",

doi = "10.1007/3-540-45540-x_11",

language = "אנגלית",

isbn = "3540428178",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "188--205",

editor = "Feitelson, {Dror G.} and Larry Rudolph",

booktitle = "Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers",

address = "גרמניה",

}

Feitelson, DG 2001, Metrics for parallel job scheduling and their convergence. in DG Feitelson & L Rudolph (eds), Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 2221, Springer Verlag, pp. 188-205, 7th International Workshop on Job Scheduling Strategies for Parallel Processing, JSSPP 2001, Cambridge, United States, 16/06/01. https://doi.org/10.1007/3-540-45540-x_11

Metrics for parallel job scheduling and their convergence. / Feitelson, Dror G.
Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers. ed. / Dror G. Feitelson; Larry Rudolph. Springer Verlag, 2001. p. 188-205 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2221).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Metrics for parallel job scheduling and their convergence

AU - Feitelson, Dror G.

N1 - Publisher Copyright: © Springer-Verlag Berlin Heidelberg 2001.

PY - 2001

Y1 - 2001

N2 - 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.

AB - 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.

UR - http://www.scopus.com/inward/record.url?scp=84896875210&partnerID=8YFLogxK

U2 - 10.1007/3-540-45540-x_11

DO - 10.1007/3-540-45540-x_11

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:84896875210

SN - 3540428178

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 188

EP - 205

BT - Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers

A2 - Feitelson, Dror G.

A2 - Rudolph, Larry

PB - Springer Verlag

T2 - 7th International Workshop on Job Scheduling Strategies for Parallel Processing, JSSPP 2001

Y2 - 16 June 2001 through 16 June 2001

ER -

Feitelson DG. Metrics for parallel job scheduling and their convergence. In Feitelson DG, Rudolph L, editors, Job Scheduling Strategies for Parallel Processing - 7th International Workshop, JSSPP 2001, Revised Papers. Springer Verlag. 2001. p. 188-205. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/3-540-45540-x_11

Metrics for parallel job scheduling and their convergence

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this