Concurrency in Distributed Systems

When multiple instances operate on shared state simultaneously, two core problems arise:

• Race condition — concurrent read-modify-write on the same resource; one operation silently overwrites the other
• Double processing — the same event or message is processed more than once (e.g. after a retry)

Strategies

• Locking — one writer at a time via optimistic or pessimistic locks
• Idempotency — safe retries; executing the same operation multiple times produces the same result