Distributed Systems
Consistency, consensus, replication, partitioning and failure — how a system stays correct when the network, the clocks and the nodes all misbehave.
1. Consistency Models
CAP and PACELC, the linearizability-to-eventual spectrum, session guarantees, and why the C in ACID is not the C in CAP.
2. Consensus & Coordination
Why agreement is hard, how Raft actually works, majority quorums and split-brain, and when you truly need a distributed lock.
3. Replication
Leader-follower, multi-leader and leaderless topologies, sync vs async durability, quorum reads and writes, replication lag and conflict resolution.
4. Partitioning & Sharding
Range vs hash partitioning, consistent hashing and virtual nodes, hot shards and skew, secondary indexes, rebalancing and request routing.
5. Time, Failure & Delivery
Failure detectors and heartbeats, clock skew and logical clocks, delivery semantics and idempotency, and the retry patterns that prevent cascading failure.