Performance analysis for parallel hash join algorithms based on DSVM and message passing

Qiang Fang*, Guoren Wang, Feng Ye, Ge Yu

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

DSVM is a new way to support parallel computing and distributed processing in NOW environment. It is different from traditional method of message passing. Three parallel hash join algorithms using these two ideas were implemented in the distributed object-oriented database system Shusse-Uo; Special test database and three different loads were designed. The performance of parallel hash join algorithms based on DSVM and message passing was analyzed in detail. The performance of algorithm based on DSVM is a little worse than that of algorithm using message passing; but it is easy for users to program in DSVM environment, so DSVM is still a valuable technology and worth further studying in parallel processing.

Original languageEnglish
Pages (from-to)583-586
Number of pages4
JournalDongbei Daxue Xuebao/Journal of Northeastern University
Volume20
Issue number6
Publication statusPublished - Dec 1999
Externally publishedYes

Fingerprint

Dive into the research topics of 'Performance analysis for parallel hash join algorithms based on DSVM and message passing'. Together they form a unique fingerprint.

Cite this