Semi-Passive Replication in the Presence of Byzantine Faults

Semi-passive replication is a variant of passive replication that does not rely on a group membership service. Défago et al. [4] defined the semi-passive replication concept in the crash fault model and described a semi-passive replication algorithm based on a lazy consensus algorithm. In this paper, we consider semipassive replication and lazy consensus for a Byzantine fault model. We present lazy Byzantine consensus algorithms for two system models: 1) a system with synchronous communication and partially synchronous processing, and 2) an asynchronous system augmented with unreliable fault detectors for Byzantine faults. We prove that our algorithms provide safety and liveness. Our algorithms are optimal in good runs, having a latency degree of 2. We describe how our algorithms can be tuned to obtain the desired levels of fault resilience or efficiency in the presence of faults. We also present optimizations to improve the performance