Average schedule length and resource selection policies on computational grids

Abstract

The computational grid provides many resources with powerful computational ability; however, we need to select appropriate resources for resolving the problem. In this paper, several simple and fast resource selection policies are presented. Under varying types of resource topologies and problems, the efficiency of these resource selection policies is compared. The contribution of this research is following. To solve the undetermined resource problem in the grid computing, the bottom level equation is modified in the list scheduling algorithm. Average Schedule Length (ASL) is presented to approximate the schedule length of the improved list scheduling method, and simulation result shows the accuracy of this approximation. The bounded number of the resources is found by performing the FindAlpha algorithm. Simulation result shows that selection policies can achieve minimal schedule length efficiently by choosing the limited number of the resources. © Springer-Verlag Berlin Heidelberg 2006.