The Research On Replica Management Strategy For Data Grid

With the rapid development and wide application of the computer network technology, many computation-intensive and data-intensive applications demand higher massive information processing and cooperative computing abilities in WAN environment. The Data Grid provides a high-performance and high-capacity parallel distributed computing platform to manage the massive data in the distributed heterogeneous WAN environment, and make the sharing of data resources of geographically dispersed organizations and the complex analysis of the large-scale data collection possible.Data replication technology has been widely applied in the areas of distributed database, mobile database, Internet and other distributed environments. In Data Grid, creating replicas of the same file in many sites can significantly reduce access latency and bandwidth consumption in accessing the huge amounts of data. Multiple replicas also avoid single point failure and improve the overall performance of the grid system.The ASHT topology combines the hierarchical and P2P topologies to organize replicas in the Data Grid. Based on ASHT topology, an adaptive and scalable dynamic replica strategy is described, which considers the network bandwidth, transfer latency, user access frequency and available storage size of the sites, and uses an evaluation function to characterize them. The result of simulation experiment shows that this replication strategy can reduce the average data access time efficiently.