Multi-Paxos

RU: Multi-Paxos

An optimization of basic Paxos for replicating a sequence of commands. A stable leader runs Phase 1 (Prepare/Promise) once to acquire leadership for a log slot range, then handles all subsequent commands with only Phase 2 (Accept/Accepted), requiring a single round trip to a quorum in steady state. Leader failure triggers a new Phase 1 from a challenger, which discovers and fills any uncommitted gaps before continuing.