Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16] [ 17] [ 18] [ 19] [ 20] [ 21] [ 22] [ 23] [ 24] [ 25]

@

Journal of Information Science and Engineering, Vol. 26 No. 6, pp. 2107-2125 (November 2010)

Adaptive Dynamic Scheduling Algorithms for Mapping Ongoing M-Tasks to PR2 Grid*

CHO-CHIN LIN, CHUN-WEI SHIH AND CHIH-HSUAN HSU
Department of Electronic Engineering
National Ilan University
Yilan, 260 Taiwan
E-mail: cclin@niu.edu.tw

Grid computing provides a platform for users easily accessing worldwide distributed resources. To meet the time limit and quality requirement posted by an application running on a grid, the computing nodes allocated to the application need to be periodically redeployed. The grid which periodically redeploys computing resources to meet the needs of an ongoing application is named a PR2 grid. An M-task is a type of workflow application which performs work stage by stage. In this paper, three algorithms MaxH-LP, MaxHrLP and MaxH-rCLP are proposed for scheduling an M-task application onto the PR2 grid. The algorithms are compared with the well known algorithms Minmin and Sufferage. Our simulation results show that the average performances achieved by MaxH-rLP and MaxHrCLP outperform Minmin and Sufferage for applications of various communication computation ratios.

Keywords: scheduling algorithm, grid computing, resource redeployment, dynamic scheduling, model

Full Text () Retrieve PDF document (201011_11.pdf)

Received December 2, 2008; revised March 26 & June 3, 2009; accepted June 30, 2009.
Communicated by Tsan-sheng Hsu.
* This research was supported by the National Science Council of Taiwan, R.O.C. under Grants No. 97-2221-E- 197-012 and 99-2221-E-197-015.