To add context, Paxos allows you to have the the same state across multiple servers without any manual master/replica failover, with allowances for a minority o

To add context, Paxos allows you to have the the same state across multiple servers without any manual master/replica failover, with allowances for a minority of the servers to go down due to upgrades or failures and with clients getting the same results no matter which of the servers they connect to. (In distributed systems this is called Consensus.) It's used in distributed datastores like Zookeeper and Exhibitor, which are used in situations where you need both consistently and as close to zero downtime as possible. It also influenced Raft, a similar but simpler algorithm to accomplish the same goals. Raft is used by etcd, which is the database that powers Kubernetes. Raft has a great website with deeper explanations at https://raft.github.io/ https://raft.github.io/