Grid computing as a dynamic domain for multi-agent systems based scheduling

Abstract

‘Grid computing’ is a recently coined terminology that captures the desire to offer computing services in a widespread, transparent and reliable manner similar to the way electrical grid systems supply electricity. Grid computers consist, respectively, of several computing elements such as processors, clusters, personal computers, data storage facilities and other peripherals that appear to the users as a single high performance computers. The usage is transparent in the sense that the users do not have to know where the resources are located. Since Grid computers consist of several interconnected computing devices, matching users and resources still remains an interesting challenge. This challenge is more demanding because the resources that are shared span across geographically independent administrative domains with different policies. The resources may appear or disappear from the grid computing pool without notice. In this paper, we review the scheduling strategies used in grid computing and identify multi-agent systems based scheduling options. We present our model that investigates how multi-agent systems can be used to support scheduling functions in a dynamic environment. Our main contribution is a more focused isolation of scheduling strategies in grid computing and their four-group categorization. We also examine the feasibility of setting up a campus grid computing services in an institution of higher learning such as a University in a developing country.