ABSTRACT

This paper deals with the resource scheduling algorithm for multi cluster hybrid grid environment. The combination of both the centralized and decentralized grid environments is collectively called as the hybrid grid environment. In this environment, each cluster has its own scheduler and cluster information system, and the whole organization is managed by one global grid scheduler with a grid information system. This mixed approach has the benefit of the shared management and administration of local and global schedulers. The local schedulers are responsible for the management of their own resources and the global scheduler manages the local schedulers. In the hybrid approach, as in the decentralized model, users submit their jobs to the appropriate cluster, and the cluster information is updated at a specific interval. The scheduler in the cluster then searches the nodes for executing the jobs in the originating cluster itself. If the number of nodes in the cluster is not satisfied, then the job will be transferred to the grid level scheduler. As in the centralized model, the grid scheduler schedules the jobs coming into it. The experimental results demonstrate that, the proposed hybrid algorithm effectively schedule the grid jobs thereby reduces total time of the jobs submitted in the grid. Also, it increases the percentage of jobs completed within the specified deadline and making the grid trustworthy.