A Study on Super-Scheduling and Data Replication Strategies in a Data Grid

In this paper, a study on the impact on performance of several super-scheduling strategies and replica optimization policies in a data grid scenario will be presented. High energy physics is a discipline that involves the execution of data-intensive jobs. Recently, computing models adopted in the Science are taking into account Grid based solutions by the European-DataGrid Project to process their data-intensive jobs. From the system point of view, Data Grids are designed around objectives such us resource utilization, job response time and overall throughput optimization. Moreover, distributed, multiple and independent resources, like computer, storage and network resources require a more complex scheduling approaches well known as super-scheduling. Study on super-scheduling and data replication strategies may reveal interesting insight into optimizing the grid cost and benefit. Simulation results about this study will be shown.