CSE 124
2017 November 28: Deeper understanding of Raft consensus

In this course, we’ve studied two major principles for building distributed systems: replicated state machines and two-phase commit. Thus far, we’ve been assuming the presence of a leader which manages these two operations, and never fails. If you’re interested in going deeper, and learning how to build a system in which the leader can fail (and a new leader can be elected), take a look at the “Raft” line of work by Ousterhout et al.

The paper is located here, and the general Raft project website is here.