Evaluating Performance Tradeoffs in a One-to-Many Peer Content Distribution Architecture

Abstract

Peer-to-Peer (P2P) content exchange has recently gained attention from both the research and industrial communities. The dynamic nature of peer networks and the resource constraints of peer hosts have introduced a number of unique technical challenges that must be addressed to make large-scale P2P content exchange applications more efficient. In this work, we expand our previous work on Pixie, an architecture that integrates one-to-many distribution of content and peer networks. Pixie uses a jukebox-style scheduling mechanism to provide a valuable data location service. Users can browse a listing of all content scheduled to be distributed across the network thus reducing search overhead. Moreover, Pixie’s use of one-to-many content distribution provides additional scalability. Our results indicate that, using Pixie, we can significantly reduce the resources required to locate and distribute content in peer networks. The properties Pixie embodies will become increasingly important as peer content exchange is extended to support more advanced, and possibly commercial applications.