Study On Replica Selection Strategy In Data Grid

In many scientific applications, Grid technologies and infrastructures facilitate distributed resource sharing and coordination in dynamic, heterogeneous multi-institutional environments. Replication of data can help enable high-throughput file transfer and scalable resource storage in scientific Grid applications that involve large data transfers. The selection of a replica can, however, significantly influence the efficiency of a replication scheme.The contribution of this paper is find the "best" replica, all we need do is to resolve Nash Equilibrium to propose a market-based strategy to replica selection for Data Grid (MSRS) based on a presupposition that all data sources are available. The goal of MSRS is to resolve replica selection problem for Data Grid efficiently and to provide foundational optimize service for large-scale data application and information disposition.Firstly a market-based model to replica selection for Data Grid is suggested, in which price, preference and utility are used to characterize the replica selection metrics. We regard replica selection as an N person games problem. Thus in order to find the "best" replica, all we need do is to resolve Nash Equilibrium. We implement the approach and evaluate it in a simulate Grid environment. Our evaluation demonstrates that the MSRS can be an efficient tool for replica selection especially in the situation that large-scale synchronous data requests occur.Another innovation of this paper is making use of user-defined "satisfaction degree" to implement replica generation automatically. Two replica placement algorithms, E-naive and E-smart, are given based on the replica clustering faced user. Through simulation experiment, the efficiency of the two algorithms has been analyzed.Finally the realization of replica selection service is discussed.