Design And Research Of Advanced Resource Class-Ification Based Grid Resource Discovery System

Grid enables the sharing, selection and aggregation of geographically distributed heterogeneous resources for solving large-scale problems in science,engineering and commerce. Resource sharing is a basic point in Grid reserch, and grid resource discovery is the premise of resource sharing in grid which concerns discovering resources in Grid to meet the requirement of applications. As resource in grid is distributed, vast in type and amount, dynamic and heterogeneous, the resource discovery should be un-centralized controlled, distributed, expand, self-organized, high-efficient, and it also should have the ability to meet the dynamic change of resource. Resource Class-ification based grid resource discovery is a decentralized gird resource discovery scheme.This paper presents an Advanced Resource Class-ification Based Grid Resource Discovery and improves on traditional methods in the following two aspacts:Organizational structure of the grid resources:(1) Central node within a resource community is dynamic selected, which ease the central node and share a single point of failure problem bottlenecks; GridSim is used for the simulation of ARC comparing with the traditional method, for multiple resource requests for the same time, ARC method shows more stable than the traditional method in success discovery rate; (2) Node joins only a limited community, which reduce the node to join unnecessary community and lead to excessive maintenance costs; For the entire network to meet a large number of continuous requests, ARC method can reduce overall network maintenance costs than traditional methods under the same success discovery rate.Resource discovery mechanism in grid:Weighted small world model is implyed to locate resource within one community, this method returns the node who can provide the maximum amount of resources to the request node, this reduces the burden of comparison and selection for the reques node. For only one resource request, ARC method returns the results with the traditional method in the same rate, but it costs less hops and give priority to node with the largest volume of request resource.