Multi-objective Functions in Grid Scheduling

Abstract

In order to fully utilize the Grid resources, an implementation of a good scheduling algorithm is greatly important. However, for a complex scheduler that aims to achieve high performance for more than one performance metrics, a suitable objective function should be carefully considered. This paper shows that a different objective function will have different affect to the Grid performance. IntroductionRecently, Grid computing have been attracting interest from researchers in the field of mathematics [8], biosciences [10], engineering [17] and others important fields. Furthermore, the increasing need for network, storage and computing resources is projected to double every 9, 12 and 18 months respectively [7]. As the increase in data size, processing complexity as well as communication technology is becoming more challenging in the Grid computing environment, issues related to performance becomes increasingly important. Achieving high-performance Grid computing requires techniques to efficiently and adaptively allocate jobs to available resources in a large scale, highly heterogeneous and dynamic environment. Therefore, it is very important for a Grid system to have a scheduler that meets both administrator and user performance expectation.