Peer-to-peer resource trading in a reliable distributed system

Abstract

Peer-to-peer architectures can be used to build a robust, fault tolerant infrastructure for important services. One example is a peer-to-peer data replication system, in which digital collections are protected from failure by being replicated at multiple peers. We argue that such community-based redundancy, in which multiple sites contribute resources to build a fault-tolerant system, is an important application of peer-to-peer networking. In such a system, there must be flexible, effective techniques for managing resource allocation. We propose data trading, a mechanism where a site acquires remote resources in the community by trading away its own local resources. We discuss the application of data trading to the data replication problem, and examine other applications of trading. A general trading infrastructure is a valuable part of a peer-to-peer, community-based redundancy system.