A new replica placement algorithm for improving the performance in CDNs

In this paper, a novel cost-based replica placement algorithm to make informed placement decisions is proposed for improving the performance of Content Distribution Networks (CDNs). From a game theoretic perspective, we design the cost function to describe various factors which to be considered when determining the optimal position of replicas. Furthermore, considering cost function and heuristic greedy algorithm, a cost-based replica placement scheme is presented for CDNs to reduce the communication latency between clients and Internet Service Providers (ISPs) and save the bandwidth consumption. Then the performance of the proposed replica placement strategy is compared with other existing methods, and the replica placement algorithms are evaluated by using both synthetic and real network topologies. Simulation results show that the cost-based placement strategy of web replicas is crucial for trade off management and computation cost for better client performance, such that it yields close to optimal performance within a factor of 82%-91% at low computation cost.