Directing requests in a large-scale Grid system based on resource categorization

This paper studies the resource discovery problem in a large-scale Grid system, in which the categorization of resources plays a crucial role. The proposed resource discovery scheme suggests that routers and resources comprise virtual organizations (VOs) within a Grid system, where a router controls locally a number of heterogeneous resources in each virtual organization. Through a simple matchmaking framework, each resource is identified as being part of a certain category, using its technical specifications (i.e. disk, and memory). The goal is to discover a suitable resource for a request, and then effectively direct this request to the specific VO that controls locally the appropriate resource. Furthermore, the proposed resource discovery scheme effectively manages the cases of requests that were directed to resources not fully suited for their satisfaction. The proposed scheme is capable of detecting these inaccuracies, so that the directing of requests occurs in a guaranteed and successful way. Finally, this paper presents the performance evaluation of the proposed resource discovery scheme by providing a number of simulation tests in large-scale Grid systems.