The Performance Evaluation of Data Caching in the Grid Environment

High latency network and large data sets produced by the applications in data grid environment can put heavy burden on network resources and cause applications to take longer times to finish. Data caching is one of the methodologies which can improve the efficiency of data usages in the distributed environment by storing popular data on the places closed to the clients. It is quite obvious that caching can reduce the bandwidth usages and minimize access latencies [1]. Although there is a study of caching in grid environment, the study focused mostly on replacement policy[2, 3, 4]. Thus, the behavior and the effect of data caching in grid environment is still relatively unknown. In this paper, we study the impacts of integrating block-based data proxy in the grid environment, we focus on the affect of the data proxy’s cache size to file access latencies, hit ratios, network bandwidth reductions and the total execution time. The experiment simulates Grid Datafarm’s data access behavior[5] and performs in NS-2[7]. The results indicate that grid caching can reduce the network bandwidth consumption and improve performance in the grid environment.