Enhanced 3-layer hierarchical replication algorithm for Data Grid

In Data Grid, large quantity of data files are generated and many users around the world need to access such data for their research and experiments. Replication is a key optimization technique to manage such a huge amount of data by replicating data in multiple grid sites. Replication goals are how to decrease bandwidth consumption, improve fault tolerance, and enhance scalability. Due to dynamic network characteristic and user behavior the main challenges are when and where to replicate data in order to achieve replication goals. Other challenges are improvement of data access efficiency since number and size of storage devices available in grid are limited while large sizes of data files are produced. In this paper, the performance of the 3-layer hierarchical replication algorithm is enhanced by placing the replica in the appropriate sites i.e. best site that has the highest number of access for that particular replica and uses an economic model for file deletion when there is not enough space for replica. The economic model is based on future value of a data file. The proposed algorithm is implemented by OptorSim, European Data Grid simulator. Experiment results show that the proposed strategy achieves better performance by minimizing the data access time and avoiding unnecessary replication.