Metadata Management Technology For Data Grid

Since data intensive computing promises dramatic progress and information processing capability has been improved gradually, there are urgent needs to expand data management capabilities of the grid and obtain the effective utilization of metadata. Therefore, establishing an effective architecture of metadata management will play an extremely important role in improving the information resource discovery, location, indexing and access.3MDG is a metadata management model for Data Grid based on dynamic balanceable placement. It provides a flexible, balanceable and self-adaptive metadata placement and maintenance platform for metadata supporters, and offers a convenient and fast metadata operated skills for metadata users, such as placement and search services. Users can rapidly find the location of physical data in order to obtain the necessary information.3MDG uses three modules and a mapping table to realize metadata management under Data Grid. The registration module supports certain functions, such as the registration and maintenance services of metadata, the registration and dissemination of metadata standards, etc. The placement module offers services of metadata location and redistribution. The cache module can be used to store hot metadata which are frequently accessed. The aim is to improve the efficiency of reading and writing metadata. Metadata location mapping table can help to describe the distributed situation of metadata that have been stored in meta-databases. In this way, system can locate the requested metadata information rapidly. Through the test of the system, the results indicate the performance of metadata management is stable.The metadata placement method uses an improved data object placement algorithm based on dynamic interval mapping. This algorithm can achieve a balanced and extendable metadata redistribution. Through the test of the algorithm, the results manifest the improved algorithm can improve the efficiency of metadata location and query function on the basis of providing balanceable and self-adaptive placement of metadata in the storage nodes.