The proposed research is primarily concerned with distributed algorithms, particularly as they relate to the performance and reliability of distributed systems. It is involved with the development of distributed algorithms that are based on the optimistic paradigm. Optimistic algorithms have been developed that particularly address the issues of rollback and concurrency in the system. The proposed algorithms for replicated data significantly reduce the latency incurred by transactions due to replica and concurrency control. It is proposed to analyze and evaluate the performance of these algorithms and to build a testbed in which these algorithms can be implemented. The goal is to develop a general framework that can be used to synthesize optimistic protocols from their centralized counterparts.
