“In Search of an Understandable Consensus Algorithm” is the paper by Diego Ongaro and John Ousterhout that introduced Raft. The extended version is published as raft.pdf on the official Raft site. The Raft site notes that a shorter version of the paper “won a Best Paper Award at the 2014 USENIX Annual Technical Conference (ATC).”
The paper’s defining feature is its stated goal. Rather than presenting consensus purely in terms of performance or theoretical minimality, the authors argued that an algorithm’s understandability matters for building correct real systems, and they designed Raft explicitly to be easier to understand than Paxos while remaining equivalent in fault-tolerance and performance.
To achieve this, the paper decomposes consensus into the relatively independent subproblems of leader election, log replication, and safety, and it reduces the number of states that must be considered by enforcing a strong form of leadership. The result is presented not only as an algorithm but as a teaching artifact, complete with safety arguments.
The paper has had outsized practical impact. It is the foundational reference for a generation of replicated systems and is paired with Ongaro’s longer Stanford PhD dissertation, which expands the algorithm into a full treatment of consensus in practice.