A service discovery model for computational grids

Computational grids (CGs) are geographically distributed aggregates of computing resources bound by a management mechanism supporting the provision of computing services; including compute cycle, business applications, and data mining and processing services. CGs span multiple management domains since their resources may be contributed by distinct providers. In this paper we present a decentralized strategy of service discovery that utilizes a selective resource state dissemination and an experience-based confidence model which provides a measure of the likelihood that a discovery request forwarded to a peer cluster would lead to a successful outcome. According to the simulation results, the proposed algorithm outperforms the random forwarding and random second-neighbor forwarding strategies respectively