Distributed consensus protocol with a coordinator

Abstract

A number of current distributed applications require the use of consensus protocols. Two well known examples are fault-tolerant applications built with process replicas in need to agree on their external behavior, and systems of replicated data where consistency has to be maintained. In this paper we rst present an already known consensus protocol which has a cost of O(n 2 ) in the number of exchanged messages, and O(n) in terms of time needed to arrive at an agreement. We then introduce a variation to this protocol which makes it linear|in the absence of failures|both in terms of exchanged messages and in time spent, and show how both consensus protocols can be used to implement reliable broadcast. Contrary to what is currently common practice, our broadcast protocols are uniform. We conclude with some possible optimizations to our protocols, and with some remarks on current and future work.