High Available Fault Tolerant Technique in Distributed Shared Memory

Abstract

distributed systems, that are based on constructing a network of heterogeneous computers, suffer from the problem of failing components during the system run time. In case of failure, the distributed applications must be restarted from the scratch. The main goal of this research is to add the dynamic failure recovery technique to the JavaSpaces server. So, the client continues its jobs while failures occur in the system. Also, the new technique in JavaSpaces is evaluated by analyzing and testing. General Terms Parallel systems, Distributed shared memory, Fault Tolerant.