Diffusive Load Balancing of Loosely-Synchronous Parallel Programs over Peer-to-Peer Networks

Abstract

The use of under-utilized Internet resources is widely recognized as a viable form of high performance computing. Sustained processing power of roughly 40T FLOPS using 4 million volunteered Internet hosts has been reported for embarrassingly parallel problems. At the same time, peer-to-peer (P2P) file sharing networks, with more than 50 million participants, have demonstrated the capacity for scale in distributed systems. This paper contributes a study of load balancing techniques for a general class of loosely-synchronous parallel algorithms when executed over a P2P network. We show that decentralized, diffusive load balancing can be effective at balancing load and is facilitated by the dynamic properties of P2P. While a moderate degree of dynamicity can benefit load balancing, significant dynamicity hinders the parallel program performance due to the need for increased load migration. To the best of our knowledge this study provides new insight into the performance of loosely-synchronous parallel programs over the Internet.