Replica placement in data grid: A multi-objective approach

Abstract

One of the challenges in data replication is to select the candidate sites to place replicas. We use the p-median and p-center models to address the replica placement problem. In our problem, the p-median model finds the locations of p candidate sites to place a replica to optimize the total (or average) response time. The p-center model hosts replicas to p candidate sites by minimizing the maximum response time among sites. A Grid environment is highly dynamic so placing a replica by considering one objective, i.e., optimize average response time or optimize maximum response time, may not be always a good choice. We propose a multi-objective model that considers the objectives of p-median and p-center simultaneously to select the candidate sites that will host replicas. Simulation results demonstrate that the multi-objective model outperforms single objective models in dynamic environments such as Data Grids. © Springer-Verlag Berlin Heidelberg 2005.