Publication

Abstract : Grid computing refers to the process of aggregating resources to resolve specific problem. Scheduling and load balancing have become an integral part of grid computing. Various literature review show that the scheduling algorithms lack in effective resource utilization and load balancing algorithms lack in identifying the appropriate resources for load allocation. The aim is to propose an Expected Mean Time (EMT) scheduling algorithm which reduces the overall makespan by making empirical calculations to determine the allocation of jobs to resources based on resource load and resource execution capabilities. The scheduling algorithm identifies the best resources for load distribution by determining the capacity of each resource in the very beginning of the processing and efficiently distributes the workload by referring the resource allocator which holds the updated status of the grid resources which are retrieved from Grid Information Server (GIS). The proposed scheduling algorithm is compared with the existing algorithm with varying loads and considering several parameters. The simulation results show that the proposed algorithm produces better makespan and improved throughput while compared with the existing classical brokers that are in use today.