Group Aware Semantic Matchmaker for Grid Resource Discovery

摘要

In large-scale Grid environments, resource discovery is made challenging by a potentially large number of resources and users and considerable heterogeneity in resource types and user requests. Resource discovery is further complicated by the dynamic variation of the number of shared resources in the system, shared resource characteristics such as availability and CPU load which vary with the time. To achieve high precision resource discovery, resources must be given well defined meaning carried by semantic information added to resource descriptions. Ontologies increasingly gain importance for grid computing to semantically describe resources and to support an improved interoperability between applications. It allows customization of the service discovery process as it provides a selection of service matches. In this paper, a framework for resource discovery based on group aware semantics is proposed. The resources are grouped according to the services offered by them and whenever there is a request for a resource the best match is returned by matchmaking locally or by forwarding the requests to the neighbor matchmakers. A flat, decentralized model is implemented so that it will thwart single point failure and increase scalability.