How do nodes agree over an unreliable network? Deep dive into the Two Generals' problem, Raft, leader election, and Byzantine fault tolerance.
Why reaching agreement over an unreliable network is mathematically impossible (FLP Impossibility), and why we settle for "probabilistic" consensus.
How systems like ZooKeeper and etcd avoid split-brain by electing a single coordinator using leases, heartbeats, and randomized timeouts.
Deep dive into Term numbers, uncommitted logs, and the commit index. Play as the leader trying to get a log entry committed to a majority of followers.
Moving beyond crash-faults to Byzantine faults (malicious actors). Practical Byzantine Fault Tolerance (PBFT) and why blockchains are incredibly slow.