A Study on Effects of Sequential Workloads on Parallel Workloads in Workstation Clusters


The Transactions of the Korea Information Processing Society (1994 ~ 2000), Vol. 5, No. 8, pp. 2137-2153, Aug. 1998
10.3745/KIPSTE.1998.5.8.2137,   PDF Download:

Abstract

In order that NOW(Network of Workstations) which is a cluster of workstations connected by a network can provide an efficient environment for both sequential programs and parallel programs, we should understand how these two kinds of programs affect the performance of each other. In this paper, we examine how the change in the sequential workloads affect the performance of parallel programs. With this purpose we implemented and experimented with a NOW simulator. In the implementation, we used synthetic workloads for sequential programs and parallel programs and we assumed coscheduling and process migration for the execution environment of parallel programs. In the experimentation we collected and analyzed data on how the number of idle workstations which is controlled by sequential programs affects the performance of parallel programs and how the performance of a parallel program is affected if several parallel programs are executed simultaneously.


Statistics
Show / Hide Statistics

Statistics (Cumulative Counts from September 1st, 2017)
Multiple requests among the same browser session are counted as one view.
If you mouse over a chart, the values of data points will be shown.


Cite this article
[IEEE Style]
C. Y. Min and S. Y. Chul, "A Study on Effects of Sequential Workloads on Parallel Workloads in Workstation Clusters," The Transactions of the Korea Information Processing Society (1994 ~ 2000), vol. 5, no. 8, pp. 2137-2153, 1998. DOI: 10.3745/KIPSTE.1998.5.8.2137.

[ACM Style]
Chang Young Min and Shim Young Chul. 1998. A Study on Effects of Sequential Workloads on Parallel Workloads in Workstation Clusters. The Transactions of the Korea Information Processing Society (1994 ~ 2000), 5, 8, (1998), 2137-2153. DOI: 10.3745/KIPSTE.1998.5.8.2137.