Distributed Systems Fundamentals

TL;DR

Distributed system: Multiple computers working together. Challenges: Network failures, clock synchronization, consensus. Consensus algorithms: Raft, Paxos (elect leader, agree on state).

Core Concepts

Challenges

Problem	Impact	Solution
Network partition	Nodes can't communicate	CAP theorem (choose CP or AP)
Partial failures	Some nodes fail, not all	Timeouts, retries, circuit breakers
Clock skew	Clocks don't match	Vector clocks, logical timestamps
Concurrency	Conflicting updates	Distributed locks, consensus

Consensus (Raft Algorithm)

Key properties:

• Majority vote: Need >50% nodes (3/5, 4/7)
• Leader election: One node coordinates writes
• Log replication: Leader replicates to followers

Distributed Locks

# Acquire lock (SET if not exists, with TTL)
locked = redis.set("lock:cron_job", "server_1", nx=True, ex=30)

if locked:
    try:
        run_cron_job()
    finally:
        redis.delete("lock:cron_job")

Challenges:

• Lock timeout: Use TTL (what if process crashes?)
• Split brain: Use consensus algorithm like Redlock

Quick Reference

CAP theorem: CP or AP (can't have both during partition)
Consensus: Raft, Paxos (leader election, agreement)
Distributed locks: Redis, etcd, ZooKeeper

---