Real-time performance improvements for distributed databases on a wide area network

An effort to improve the performance of distributed processing applications on very-high-speed wide area networks is described. The primary limiting factor in this case appears to be latency, the innate propagation delay connected with sending signals over a long distance. A method to mask the effects of latency by continuously sending large amounts of status information over an endless loop connecting the various nodes in the distributed system is proposed. Particular attention is given to the case of unbalanced transaction arrival patterns. Simulation results are presented