Failure detector-Ring Paxos-based atomic broadcast algorithm

Abstract

An atomic broadcast problem constitutes an essential component in fault-tolerant distributed systems. an atomic broadcast algorithm ensures that all processes deliver the same messages sequence. Many algorithms solving the atomic broadcast problem have been published. Recently, a new Paxos-based atomic broadcast algorithm has been proposed for clustered systems, named Ring Paxos. It inherits many of Paxos characteristics: safe under asynchronous assumptions, live under weak synchronous assumptions, and requires a majority of non-faulty processes to ensure progress. However, the proposed algorithm relies on strong assumptions, and uses the group membership for faults tolerance. Recently, a new failure detector Ring Paxos-based atomic broadcast algorithm (FD-RP) for asynchronous distributed systems have been proposed. It inherits some of Ring Paxos characteristics and uses failure detectors to tolerate failures instead of a membership service. The aim of this paper is to complete the work done in our previous work by evaluating the performance of the FD-RP protocol by simulation and compare it with the Ring Paxos protocol, results show that the proposed protocol performs better than the Ring Paxos protocol in some situation and is recommended in applications with low latencies.