Capacity and load-aware service discovery with service selection in peer-to-peer grids

In recent years, peer-to-peer (P2P) grids have emerged as a new powerful computing infrastructure which allows participating peers/end users to share their resources in an efficient manner. In P2P grids, peers/nodes act as both service providers (SPs) and also service consumers. There are a number of services available to end users in P2P grids. The use of a reciprocation mechanism in P2P grid systems is an efficient way for clustering peers for efficient service discovery and service selection. However due to the limited resources of each peer, they can only provide a subset of all possible services. Moreover, due to the highly dynamic nature of P2P grids, it is not feasible for a participating peer to decide the optimal selection of services that it should have. These factors limit the quality of service (QoS) among the participating peers. Hence, to maximize the QoS among the peers, a particular heuristic is needed which has a direct impact on the profit that the grid can provide to end users. In this paper, we propose a capacity and load-aware service selection (CLSS) and peer cache-based service discovery (PCSD) algorithms to maximize profit with minimum resource consumption in P2P grids. The problem of service discovery and service selection is formulated as a linear programming (LP) problem together with the constraints and proposed algorithms. The performance of the proposed algorithms is evaluated with respect to metrics such as total execution time, successful execution rate, service availability success, and impact of capacity on load on individual peers. The results obtained show that the performance of the proposed algorithms is better than the other algorithms with respect to these metrics.

► Capacity and load-aware service selection with peer cache-based service discovery.
► The problem is formulated as a linear programming (LP) problem.
► The results obtained show that the performances of the proposed algorithms are better.