A Novel Replication Strategy in Data Grid Environment with a Dynamic Threshold

Data Grid is a type of Grid Computing systems whichis designed to provide geographically distributed data resources to large computational problems that require mining and evaluating large amounts of data. Managing this data in a centralized location increases the data access time and hence much time is taken to execute the job. So to reduce the data access time, "Replication" is used. Data replication is known as an important optimization technique that aims to improve data access time and toutilize network and storage resources efficiently.Since the data files are very large and the Grid storagesare limited, managing replicas in storage for the purpose of more effective utilization requiresmore attention.In this paper, a novel data replication strategy, called Dynamic Hierarchical Replicationwith Threshold (DHRT) is proposed. This strategy is an enhanced version of the Dynamic Hierarchical Replication (DHR)strategy that uses a new threshold for characterizing the number of appropriate sites for replication. Appropriate sites have the higher number of access for that particular replica from other sites. It also minimizes access latency by selectingthe best replica when various sites hold replicas. The proposed replica selection strategy selects thebest replica location for the users' running jobs by considering the replica requests that are waite in thestorage and number of stored file. The simulated results with OptorSim, i.e. European Data Grid simulatorshow that the DHRT strategy gives better performance compared to the other algorithms and preventsthe unnecessary creation of replicas which leads to efficient storage usage.