Knowledge trunk
Distributed systems
Systems where communication, partial failure, duplicate work, and stale knowledge become ordinary conditions.
updated 2026-08-25
Once work crosses a machine or process boundary, failure stops being a single clean event. Messages can be delayed, duplicated, reordered, or lost; a caller can stop waiting while the callee continues; two replicas can each have a plausible but different view of reality.
Start with Retries and timeouts, then Idempotency. Replication and consistency widens the question from one request to several durable copies of state.
This trunk links constantly into transactions, cancellation, authority, and tail latency.
Local nodes