A Novel Job Scheduling Algorithm for Improving Data Grid's Performance

Abstract

Data Grid is a geographically distributed environment that deals with large-scale data-intensive problems. The main problems in data grid are job scheduling and data management. Generally, job scheduling in Grid has been studied from the perspective of computational Grid. In Data Grid, effective scheduling policy should consider both computational and data storage resources. In this paper a new job scheduling algorithm, called Combine Scheduling Strategy (CSS) is proposed that considers number of jobs waiting in the queue, location of required data and the computing capacity of sites. Scheduling cannot be effective unless to combine it with replication. Therefore, we have discussed various strategies in scheduling and replica optimization. Simulation results demonstrate that CSS gives better performance compared to the other algorithms.