On Server Dimensioning for Hybrid P2P Content Distribution Networks

One of the key questions arising in the dimensioning of a hybrid P2P content distribution system is that of the required infrastructure support in terms of server bandwidth. We develop and propose simple mathematical models as a methodology for analyzing and dimensioning hybrid peer-to-peer content distribution networks. We first use a deterministic fluid model to capture the essential peer and server dynamics within a single swarm, and subsequently derive a stochastic fluid model to capture the dynamics in the case of multiple swarms, i.e., concurrent swarms of a number of content objects. Based on the models, we derive solutions for estimating the server capacity required to support a single swarm as well as a number of concurrent file swarms at a given level of service quality. Numerical results demonstrate how a hybrid P2P approach can yield substantial performance gains and capacity savings compared to a pure client/server system, with churn rate and upload bandwidth being critical factors. Compared to a pure peer- to-peer scenario, the hybrid approach can dramatically boost the performance and improve reliability.