A study on performance of dynamic file replication algorithms for real-time file access in Data Grids

Abstract

Real-time Grid applications are emerging in many disciplines of science and engineering. In order to run these applications while meeting the associated real-time constraints with them, the Grid infrastructure should be designed to respect these constraints and allocate its Computing, networking, storage, and the other resources accordingly. Furthermore, these applications involve a large number of data intensive jobs and require to access terabytes of data in real-time. On the other hand, a variety of dynamic file replication algorithms were proposed for the best-effort Data Grid environments in an attempt to decrease job completion times and save network bandwidth. Until now, there is no study in the literature which tries to elaborate on the real-time performance of these dynamic file replication algorithms. Based on this motivation, in this study, the performance of eight dynamic replication algorithms are evaluated under various Data Grid settings. For this evaluation, a process oriented and discrete-event driven simulator called DGridSim is developed. A detailed set of simulation studies are conducted using DGridSim and the results obtained are presented to reveal the real-time performance of the dynamic file replication algorithms