Replica Placement in Multi-tier Data Grid

Replication is used in distributed and pervasive computing systems to enhance availability by placing needed data close to the users so that user perceived response time is minimized. In multi-tier data grid environment, replicas are usually placed in either top-down or bottom-up way. We believe that availability of data can be greatly enhanced by using both methods in a coherent way. Our proposed scheme, two-way replication, identifies and places the popular data to the requesting sites in a bottom-up fashion. Using top-down way, it places the less popular data a tier below from the data grid root. In this system, requests for data can be serviced from the siblings as well as the parent sites of the requesting clients. Experiment results reveal that two-way replication fulfills the availability needs of users with less number of replicas and more economic storage utilization as compared to fast spread technique.