Mahi-Mahi: Low-Latency Asynchronous BFT DAG-Based Consensus

We present Mahi-Mahi, the first asynchronous BFT consensus protocol that achieves sub-second latency in a wide-area network setting while processing over 100,000 transactions per second. Mahi-Mahi achieves such high performance by leveraging an uncertified structured Directed Acyclic Graph (DAG) to forgo explicit certification. This reduces the number of messages required to commit and the CPU overhead for certificate verification, significantly. Mahi-Mahi introduces a novel commit rule that enables committing multiple blocks in each asynchronous DAG round. Mahi-Mahi can be parametrized either with a 5 network hops commit delay, maximizing the commit probability under a continuously active asynchronous adversary, or with a 4 network hops commit delay, reducing latency under a more moderate and realistic asynchronous adversary. We demonstrate safety and liveness of Mahi-Mahi in a Byzantine context for all of these parametrizations. Finally, we evaluate Mahi-Mahi in a geo-replicated setting and compare its performance to state-of-the-art asynchronous consensus protocols, showcasing Mahi-Mahi’s significantly lower latency.