Increasing Reliability of Communication in Large Scale Distributed Systems

This paper discusses the problem of reliable communication in large scale networks. More precisely, we describe the problem of {\it non-transitive} communication in a large scale distributed system due to link failure which leads to {\it partial} partitions. A definition for {\it partial} partition in contrast to a {\it total} partition is given. The solution to mask these {\it partial} partitions and as a consequence providing {\it transitive} communication, consists in using every possible site in the distributed system to reroute messages. A communication layer using rerouting of messsages can be much more reliable.