Parsimonious asynchronous byzantine-fault-tolerant atomic broadcast

• Authors: HariGovind V. Ramasamy;Christian Cachin
• Venue: OPODIS'05 Proceedings of the 9th international conference on Principles of Distributed Systems
• Year: 2005

Atomic broadcast is a communication primitive that allows a group of n parties to deliver a common sequence of payload messages despite the failure of some parties. We address the problem of asynchronous atomic broadcast when up to t n/3 parties may exhibit Byzantine behavior. We provide the first protocol with an amortized expected message complexity of $\mathcal{O}(n)$ per delivered payload. The most efficient previous solutions are the BFT protocol by Castro and Liskov and the KS protocol by Kursawe and Shoup, both of which have message complexity $\mathcal{O}(n^2)$. Like the BFT and KS protocols, our protocol is optimistic and uses inexpensive mechanisms during periods when no faults occur; when network instability or faults are detected, it switches to a more expensive recovery mode. The key idea of our solution is to replace reliable broadcast in the KS protocol by consistent broadcast, which reduces the message complexity from $\mathcal{O}(n^2)$ to $\mathcal{O}(n)$ in the optimistic mode. But since consistent broadcast provides weaker guarantees than reliable broadcast, our recovery mode incorporates novel techniques to ensure that safety and liveness are always satisfied.