A Fault Recovery Scheme for P2P Metacomputers

Despite the leaps and bounds made by the P2P research field in the last few years, the benefit of this innovation has been constrained to a few areas; search and file-sharing and storage to name a few. In particular, this innovation has had little significant impact in the field of distributed computing. There are several obstacles to be overcome in the development of any distributed computer, most notably: scalability, fault tolerance, security and load balancing. The difficulty of these is compounded in the dynamic, decentralized environment which characterizes the P2P arena. This paper presents a method of recovering from faults which exploits the distributed hash table functionality provided by modern overlay networks. Its effectiveness is evaluated experimentally using a proof of concept P2P distributed computer. It is hoped that by providing a solution to one of the obstacles, global, decentralized, dependable distributed computers will be one step closer to reality.