The performance evaluation and analysis of caching in data grid system

Abstract

Data-intensive applications in the data grid environment usually transfer large amount of data, which can cause high network consumption and degrade the overall performance of the network. Caching is one of the techniques that can reduce network bandwidth usage by storing data likely to be used in the near future in its local storage. This technique is very successful, especially in the Internet. However, the conventional cache used in WWW can not be applied completely into the grid environment because of the differences between Web and Data Grid{7f2019}s behaviors, for example, data size, network consumption. This thesis proposes the block-based data caching model for data grid environment by dividing data file in many data blocks which can achieve better space utilization which leads to better overall performance. The experiment conduct by simulation using grid extended NS-2. The result indicates that data caching, especially block-based data caching, can provide benefits for the data grid environment which can be further developed into real implementation and can fine-tune for different environments.