Resource based optimized decentralized grid scheduling algorithm

Abstract

Peer to peer (P2P) grid system has good potential for decentralized grid scheduling. The existing P2P grid resource management algorithms allow detection of resources after generation of task. Recently, few new P2P grid resource management schemes proposed discovery of resources before task is generated. These schemes simply shortlist best possible subset of grid resources on basis of one or many overlays. In this paper, we suggest that these algorithms have potential to include the step of organizing shortlisted resources in non-increasing order on basis of optimization criteria. Addition of this step will reduce time to find out schedule of DAG's interdependent tasks on fully decentralized P2P grid. An optimization criterion is based on computation capability and communication cost of grid resources. © 2012 Springer-Verlag GmbH.