A load balancing replica placement strategy in Data Grid

Replication is used in data grid to enhance data availability and fault tolerance. The random burst of data requests may increase the frequency of dynamic replication thus putting system nodes under heavy storage load and delaying execution of jobs. We propose a load balancing replication strategy, Fair-Share Replication(FSR) that takes into account both the access load and storage load of the replica servers before placing a replica. We use P2P concepts in multi-tier data grid with unique path for data search, and leverage the neighborhood of replica servers in replication process. A sibling node of a candidate replica server is selected for replica placement if its access and storage loads are less than the candidate. Experiment results show that FSR performs better than fast spread replication using random access pattern.