Design And Implementation Of Distributed Grid Resource Discovery System

Grid is a basic resource discovery service. When a description of the required resources is given, Grid Resource Discovery will return the location of resources which match with the description. The current Grid Resource Discovery mechanism is generally based on either centralized or distributed models. This paper proposes a new resource discovery model that can alleviate the deficiency of the centralized and non-structured P2P (Peer to Peer) resource discovery mechanism in Grid environment. The model is based on semi-structured index-P2P method and has hybrid three-layer architecture. The top layer consists of indexing peers that forms a structured P2P network for providing publishing and searching services to other peers. The middle layer consists of service peers that can join and leave the P2P network freely. The bottom layer provides Grid services by normal peers. Performance of the model in highly dynamic environment is analyzed and tested.Distributed hash algorithm is used to solve the structural problems of distributed storage by Distributed Hash Tables. The core idea of distributed hash algorithm is: the distributed storage of object bases on the Hash Key which gets the key of stored object by hash operation. This paper designs distributed hash algorithm of the top layer. The algorithm locates resource by mapping resource keyword to the indexing peers. And the query of algorithm is expanded to improve the inquired efficiency of indexing peers.In this paper, the construction of model and resource inquiry are finally designed and implemented. And the mathematical check of the model superiority is also completed. As a result, it is shown that the searching, routing, indexing and query transmitting ability of the model are greatly improved.