Subway: Peer-to-Peer Clustering Of Clients for Web Proxy

Many cooperated web cache systems and protocols have been proposed. But, these systems need the expensive resources, such as core-link bandwidth and proxy cpu or storage, and need administrative cost to achive enough performance when the client population increases. This paper suggests a self-organizable caching system, called Subway, which is scalable and fault tolerant, with a distributed peer-to-peer storage as a backup storage. Subway reduces consumption of expensive resources by using resources of participant clients, which use the service of a central proxy. We examine the performance of Subway via trace driven