Skip to Main Content
In data grid, it is an important research filed to complete interoperability of data. In the mean time, share of data also becomes the crucial problem. Data replication, as a solved solution of data share, goes into more and more vital. A strategy called replication strategy based on clustering analysis (RSCA) is proposed, which confirms the correlation among the data files accessed according to the access history of users. And then, through clustering analysis operation obtains the correlative files sets related to the access habits of users. At the same time, it produces the data files replica on the basis of those sets, which achieves the aim of prefetching and buffering data. The experimental results show that RSCA is effective and available. Contrast to other dynamic replication strategies, it has reduced not only the average response time of client nodes, but also those of the bandwidth consumption.