Distributed Systems

Why distributed systems are fundamentally harder than single-node ones — partial failures, network uncertainty, and no shared clock.

What happens when nodes can't talk. Split brain, quorum decisions, and why partition handling defines your entire consistency strategy.

Consistency vs availability when a partition happens. CP vs AP — and why every distributed system is forced to choose during a split.

Extends CAP to cover normal operation. Even without a partition, you still trade latency against consistency — PACELC names that trade-off.

Strong, eventual, causal, and monotonic. What "up to date" means in a distributed system and when each model is the right call.

Sync vs async replication, replication lag, failover, and multi-primary — how data stays consistent across nodes.

Shard keys, consistent hashing, cross-shard joins, resharding, and the over-sharding trap that kills performance.

Raft leader election, log replication, Paxos phases, ZooKeeper, and Redis distributed locks — how nodes agree on a single value.

Clock drift, NTP, Lamport clocks, vector clocks, and TrueTime — how distributed systems reason about time without a shared clock.

Conflict-free replicated data types — the math that lets replicas merge without coordination, and why locks fail at global scale.

Heartbeats, gossip protocol, and the phi accrual failure detector — how nodes know when other nodes are dead.

Hash trees for efficient data comparison across replicas. Anti-entropy repair and how Dynamo-style systems detect divergence.

etcd, leases, TTL-based fencing tokens, and the difference between distributed locks and job tracking.