Reversibility in Message-Passing Systems

Reversibility in computation refers to the quality of computational processes being time-reversible, meaning that that the dynamics of the process remain well-defined when the sequence of time-states is reversed. This report surveys the implementation of reversibility in software of distributed concurrent systems that rely on message passing as the method of communication. More specifically, two application domains are explored: employing reversibility to achieve fault-tolerance in concurrent systems, and implementing reversible debugging in message-passing systems, by looking at existing implementations, and exploring a possibility of implementing a causal-consistent reversible debugger for the Message Passing Interface. Keywords––reversibility, message-passing interface, debugging, fault-tolerance, parallel programming