tl;dr - A few more Paxos papers have been produced since the original post, I do the usual writeup. Main benefit/focuses this time look to be centered around increasing throughput by taking load off of the leader node(s) (and assigning the extra work to existing or new nodes). Multi-part blog post alert This is (now) a multi-part blog-post! Part 1 - Paxosmon: Gotta consensus them all Part 2 - Paxosmon 2: The Journey Continues (this post) What’s new?
