The dead letter queue and why you need one

In any system that processes a stream of work, some items will fail — bad data, a temporary outage, an unexpected case. The critical question is what happens to those failures. Without a plan, they vanish, and you discover the loss only when someone asks where their thing went.

Failures need a destination

A dead letter queue is simply a designated place where items that couldn't be processed go to wait, instead of disappearing. It turns silent data loss into a visible, recoverable pile you can inspect, fix, and reprocess. The name is grim; the function is a safety net.

Why it matters

Without it, a single malformed record can either crash the whole pipeline or get dropped without a trace — both bad. With it, the good work keeps flowing, the failures are captured for review, and nothing is lost. It's the difference between a pipeline you trust and one you cross your fingers over.

Every pipeline drops some items. A dead letter queue is the difference between catching them and losing them silently.