AGENT BASED OPTIMIZED RÉPLICA MANAGEMENT IN DATA GRIDS

Abstract

Sharing of data in a distributed environment such as grid leads to various design issues such. One of the effective measures for accessing the distributed data is réplication. This paper addresses réplica management issues such as availability, placement and consistency. Proper management of réplicas in a distributed grid environment improves the accessibility of updated réplica. A strategy namely, Réplica Creation, Placement and Consistency of data (RCPC) is proposed. RCPC dynamically creates an optimal number of réplicas and places them on nodes having minimum placement cost. An agent- based adaptive threshold approach is used efficiently for maintaining consistency among the réplicated data. RCPC is simulated using Optorsim simulator and the results are evaluated in terms of job execution time, the effectiveness of network usage, hit ratio, transfer rate, storage utilization and computing usage.