Message Queue
Also known as:
Message Broker
Queue
Definition
A message queue is middleware that lets services talk asynchronously by passing messages through a broker instead of calling each other directly. It decouples producers from consumers, absorbs traffic spikes by buffering, and enables background processing. Delivery is usually at-least-once, so consumers need to be idempotent, and failed messages are parked in a dead letter queue for later inspection.
Key Takeaways
- Queues decouple producers and consumers, smooth out spikes, and let slow work happen asynchronously.
- Point-to-point delivers a message to one consumer; pub/sub broadcasts it to every subscriber.
- Most brokers are at-least-once, so duplicates happen and consumers must be idempotent.
- Dead letter queues capture messages that repeatedly fail so they do not block the main flow.
How It Works
- A producer publishes a message to a queue or topic on the broker.
- The broker stores the message durably until a consumer is ready.
- A consumer pulls the message, processes it, and acknowledges it so the broker can drop it.
- If processing fails or is not acknowledged in time, the broker redelivers, and after repeated failures routes it to a dead letter queue.
Where It Is Used
- Kafka is high-throughput streaming with replayable logs; RabbitMQ does rich routing and request-reply; SQS is simple and AWS-native.
- Events from the transactional outbox flow to consumers through a queue.
- Background jobs, email sending, and CQRS projections are all common queue consumers.