Delivery

Messages that keep their meaning across teams and systems.

Retry a request without repeating its effect

A caller can lose the response after a server has already completed an operation. Retrying is then ambiguous: the first attempt may have succeeded. Idempotency gives repeated attempts a defined relationship instead of leaving duplicates to later cleanup.

Delivery
Give a recovering service room to recover

Retries can amplify an outage when every client sends another request at the same instant. Backoff spreads attempts over time. It belongs with a maximum retry budget, otherwise a failed operation can occupy resources long after it remains useful.

Delivery
Define what an acknowledgement promises

Accepted, stored, processed and delivered describe different states. A positive acknowledgement is only useful when both sides agree which state it confirms. Confusing receipt with completion can make a dashboard look successful while work is still pending.

Delivery
Keep failed messages available for diagnosis

A repeatedly failing message should not prevent unrelated work from progressing. A dead-letter queue separates persistent failures from the normal flow while retaining enough information to understand them. It is a work queue for recovery, not permanent storage.

Delivery
Let a slow consumer signal its limits

If producers can send work faster than consumers can finish it, a queue keeps growing. Backpressure makes the imbalance visible and limits incoming work. Extra buffering may absorb a short burst, but it cannot solve sustained overload.

Delivery

Search the library