Comparing Distributed Consensus Algorithms

Peter Urban and Andre Schiper Comparing Distributed Consensus Algorithms Protocols that solve agreement problems are essential building blocks for fault tolerant distributed systems. While many protocols have been published, little has been done to analyze their performance, especially the performance of their fault tolerance mechanisms. In this paper, we compare two consensus algorithms with different communication schemes: one is centralized and the other decentralized. The elements of the simulation study form a generic methodology for evaluating consensus algorithms. The results show that the centralized algorithm performs better in some environments, in spite of the fact that the decentralized algorithm finishes in fewer communication steps. The reason is that it generates less contention. Keywords: simulation tools and techniques, Java-based simulation, distributed consensus, benchmarks