Queue Backpressure Lab: Keep Workers Alive During a Traffic Spike

A queue is supposed to protect a system, but it can also hide disaster until everything is behind. Participants operate a small worker system through a traffic spike, slow downstream API, retry storm, and poison message.

The lab covers concurrency limits, backoff, dead-letter queues, visibility timeouts, idempotent jobs, queue depth alerts, and graceful degradation. Attendees leave understanding how to keep background work reliable when demand is not polite.