Cost Saving Replication Technique in Data Grid

Abstract

Replication is a useful technique for distributed database systems. An existing popular technique that has been used before are Read One-write-All Monitoring Synchronization Transactions System (ROWA-MSTS) and Hierarchical Replication Scheme (HRS). Nevertheless, these techniques have their flaws in terms of replication time taken. Subsequently, ROWA and HRS take more executing time in order to complete a transaction. It is because they have to copy their data to all sites. This research, Binary Voting Fragmented Database Replication (BVFDR) model, the some-data-to-some-sites technique is proposed. In order to handle transactions in the systems, BVFDR considers the neighbouring servers binary vote assignment to its logical grid structure on fragmented data copies. Therefore, it reduces storage capacity. An experiment has been carried out in three replicated servers. The results have been compared with existing techniques such ROWA and HRS. The result shows that BVFDR is able to preserve data consistency and outperformed in terms of time taken for a complete transaction compare to existing techniques. Overall, BVFDR able to handle fragmented data replication as well as capable to manage transaction management in distributed database environment by maintaining consistency of the data through the synchronization approach.