Multi-agent systems scheduler and open grid computing standards

Abstract

Grid computers are integrated environments of software and hardware that are inter-connected and allow sharing of computing resources. A user of a grid computer has the impression of working with a single powerful computer. The development and realization of grid computers still remains a challenging task. This challenge is biased towards a scheduling task problem in a dynamic setting. We approach this complex problem by using multi-agent systems. In this paper we examine the nature of the scheduling problem in grid computing domain and show that it is an on demand decentralized resource allocation problem. We outline our model of a scheduler that is based on multi-agent systems. We highlight the interactions between agents and rely on Vickrey auction mechanism for a decentralized truth revealing strategy. We give an overview of open grid computing standards, from Global Grid Forum. We use these open grid computing standards to briefly critique our model, pointing out the similarities and differences. Our contributions in this paper include our proposed scheduling model based on multi-agent systems. We have also reviewed some market based approaches to the resource allocation problem for decentralized contexts. We note that increased reliance on agents and Web-services may one day lead to a situation where the grid computer is the Internet and the Internet is the grid computer.