Improving response time in cluster-based Web servers through coscheduling

Jin Ha Kim, Gyu Sang Choi, Deniz Ersoz, Chita R. Das

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Citations (Scopus)

Abstract

In this paper, we investigate the feasibility of minimizing the response time of a Web server by exploiting the advantages of both user-level communication and coscheduling. We, thus, propose a coscheduled server model, based on the PRESS design where the remote cache accesses can be coscheduled on different nodes to reduce the response time. We experiment this concept using two known coscheduling techniques, called Dynamic Coscheduling (DCS) and DCS with immediate blocking. Extensive simulation of four server models (PRESS over TCP/IP, PRESS over VIA, coscheduled PRESS model with DCS, and with DCS and blocking) using 16-node and 32-node cluster configurations indicates that the average response time of a distributed server can be minimized significantly by coscheduling the communicating processes. The use of the DCS scheme reduced the average latency up to 80%, on an average 40%, compared to the PRESS over VIA model that uses only user-level communication, and by order of magnitude compared to the TCP/IP model. The throughput of the three user-level communication models is at least 25% better compared to the PRESS over TCP/IP model. Experiments with varying file size and cache size also confirmed the advantage of using a coscheduling mechanism for improving the response time behavior.

Original languageEnglish (US)
Title of host publicationProceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)
Pages1237-1246
Number of pages10
StatePublished - Dec 1 2004
EventProceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM) - Santa Fe, NM, United States
Duration: Apr 26 2004Apr 30 2004

Publication series

NameProceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)
Volume18

Other

OtherProceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)
CountryUnited States
CitySanta Fe, NM
Period4/26/044/30/04

Fingerprint

Response time (computer systems)
Servers
Communication
Experiments
Throughput

All Science Journal Classification (ASJC) codes

  • Engineering(all)

Cite this

Kim, J. H., Choi, G. S., Ersoz, D., & Das, C. R. (2004). Improving response time in cluster-based Web servers through coscheduling. In Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM) (pp. 1237-1246). (Proceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM); Vol. 18).
Kim, Jin Ha ; Choi, Gyu Sang ; Ersoz, Deniz ; Das, Chita R. / Improving response time in cluster-based Web servers through coscheduling. Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM). 2004. pp. 1237-1246 (Proceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)).
@inproceedings{23e9a1f946014cdfb9941ab94c1b49c0,
title = "Improving response time in cluster-based Web servers through coscheduling",
abstract = "In this paper, we investigate the feasibility of minimizing the response time of a Web server by exploiting the advantages of both user-level communication and coscheduling. We, thus, propose a coscheduled server model, based on the PRESS design where the remote cache accesses can be coscheduled on different nodes to reduce the response time. We experiment this concept using two known coscheduling techniques, called Dynamic Coscheduling (DCS) and DCS with immediate blocking. Extensive simulation of four server models (PRESS over TCP/IP, PRESS over VIA, coscheduled PRESS model with DCS, and with DCS and blocking) using 16-node and 32-node cluster configurations indicates that the average response time of a distributed server can be minimized significantly by coscheduling the communicating processes. The use of the DCS scheme reduced the average latency up to 80{\%}, on an average 40{\%}, compared to the PRESS over VIA model that uses only user-level communication, and by order of magnitude compared to the TCP/IP model. The throughput of the three user-level communication models is at least 25{\%} better compared to the PRESS over TCP/IP model. Experiments with varying file size and cache size also confirmed the advantage of using a coscheduling mechanism for improving the response time behavior.",
author = "Kim, {Jin Ha} and Choi, {Gyu Sang} and Deniz Ersoz and Das, {Chita R.}",
year = "2004",
month = "12",
day = "1",
language = "English (US)",
isbn = "0769521320",
series = "Proceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)",
pages = "1237--1246",
booktitle = "Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)",

}

Kim, JH, Choi, GS, Ersoz, D & Das, CR 2004, Improving response time in cluster-based Web servers through coscheduling. in Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM). Proceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM), vol. 18, pp. 1237-1246, Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM), Santa Fe, NM, United States, 4/26/04.

Improving response time in cluster-based Web servers through coscheduling. / Kim, Jin Ha; Choi, Gyu Sang; Ersoz, Deniz; Das, Chita R.

Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM). 2004. p. 1237-1246 (Proceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM); Vol. 18).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Improving response time in cluster-based Web servers through coscheduling

AU - Kim, Jin Ha

AU - Choi, Gyu Sang

AU - Ersoz, Deniz

AU - Das, Chita R.

PY - 2004/12/1

Y1 - 2004/12/1

N2 - In this paper, we investigate the feasibility of minimizing the response time of a Web server by exploiting the advantages of both user-level communication and coscheduling. We, thus, propose a coscheduled server model, based on the PRESS design where the remote cache accesses can be coscheduled on different nodes to reduce the response time. We experiment this concept using two known coscheduling techniques, called Dynamic Coscheduling (DCS) and DCS with immediate blocking. Extensive simulation of four server models (PRESS over TCP/IP, PRESS over VIA, coscheduled PRESS model with DCS, and with DCS and blocking) using 16-node and 32-node cluster configurations indicates that the average response time of a distributed server can be minimized significantly by coscheduling the communicating processes. The use of the DCS scheme reduced the average latency up to 80%, on an average 40%, compared to the PRESS over VIA model that uses only user-level communication, and by order of magnitude compared to the TCP/IP model. The throughput of the three user-level communication models is at least 25% better compared to the PRESS over TCP/IP model. Experiments with varying file size and cache size also confirmed the advantage of using a coscheduling mechanism for improving the response time behavior.

AB - In this paper, we investigate the feasibility of minimizing the response time of a Web server by exploiting the advantages of both user-level communication and coscheduling. We, thus, propose a coscheduled server model, based on the PRESS design where the remote cache accesses can be coscheduled on different nodes to reduce the response time. We experiment this concept using two known coscheduling techniques, called Dynamic Coscheduling (DCS) and DCS with immediate blocking. Extensive simulation of four server models (PRESS over TCP/IP, PRESS over VIA, coscheduled PRESS model with DCS, and with DCS and blocking) using 16-node and 32-node cluster configurations indicates that the average response time of a distributed server can be minimized significantly by coscheduling the communicating processes. The use of the DCS scheme reduced the average latency up to 80%, on an average 40%, compared to the PRESS over VIA model that uses only user-level communication, and by order of magnitude compared to the TCP/IP model. The throughput of the three user-level communication models is at least 25% better compared to the PRESS over TCP/IP model. Experiments with varying file size and cache size also confirmed the advantage of using a coscheduling mechanism for improving the response time behavior.

UR - http://www.scopus.com/inward/record.url?scp=12444343065&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=12444343065&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:12444343065

SN - 0769521320

SN - 9780769521329

T3 - Proceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)

SP - 1237

EP - 1246

BT - Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)

ER -

Kim JH, Choi GS, Ersoz D, Das CR. Improving response time in cluster-based Web servers through coscheduling. In Proceedings - 18th International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM). 2004. p. 1237-1246. (Proceedings - International Parallel and Distributed Processing Symposium, IPDPS 2004 (Abstracts and CD-ROM)).