Fault tolerance management in collaborative systems: Performance comparison of consensus algorithms

It is well known that consensus algorithms are fundamental building blocks for fault tolerant distributed systems, including distributed collaborative systems. In the literature of consensus, many algorithms were proposed to solve this problem in different system models but not a lot of efforts were made to analyze their performance. This paper compares the performance of four of the most famous consensus algorithms in asynchronous distributed systems with the crash-stop failure model. We use simulation and we conduct our experiments in failure free situations and in the presence of processes crashes. Our results give a global idea of the performance of these algorithms and help us understand how to choose the most suitable consensus algorithm for each situation.