Concepts

Every SQS concept, explained clearly.

Each concept is a focused reading page — definition, parameters, edge cases, and gotchas — distilled from the AWS SQS Developer Guide. When a concept has a matching Failure Lab, you'll see a link to try it.

Fundamentals

Standard Queues

Nearly unlimited throughput, best-effort ordering, at-least-once delivery.

5 sections · 3 gotchasread

FIFO Queues

Strict ordering and exactly-once processing — per message group.

6 sections · 3 gotchasread

Short vs Long Polling

Long polling cuts empty-receive cost and reduces latency.

5 sections · 3 gotchasread

Reliability

Visibility Timeout

How long a message is hidden after a consumer receives it.

6 sections · 3 gotchasread

Retries & maxReceiveCount

How SQS retries a message until it succeeds — or gives up.

5 sections · 3 gotchasread

Dead Letter Queue (DLQ)

Quarantine for poison messages so they don't take down healthy traffic.

5 sections · 4 gotchasread

Redrive Policies

Send messages from a DLQ back to the source queue once the bug is fixed.

5 sections · 4 gotchasread

Idempotency

Processing the same message twice must yield the same result.

5 sections · 3 gotchasread

Ordering & dedup

Message Group ID (FIFO)

The partition key that controls per-group ordering and parallelism.

5 sections · 3 gotchasread

Deduplication ID (FIFO)

Drop duplicate sends inside a 5-minute window.

4 sections · 3 gotchasread

Exactly-once Semantics

FIFO + deduplication + idempotent consumers = effectively exactly-once.

4 sections · 2 gotchasread

Timing

Delay Queues

Hide all new messages from consumers for N seconds.

4 sections · 3 gotchasread

Message Timers

Per-message DelaySeconds for one-off scheduled work.

4 sections · 2 gotchasread

Message Retention

How long an unprocessed message lives in the queue.

4 sections · 3 gotchasread

Scale & performance

Batching

Send and receive up to 10 messages per API call.

5 sections · 3 gotchasread

Scaling Consumers

Horizontally scale on queue depth, not on CPU.

5 sections · 3 gotchasread

Backpressure

What happens when producers outpace consumers.

4 sections · 3 gotchasread

Queue Throughput

Standard is effectively unlimited. FIFO has hard limits.

4 sections · 3 gotchasread

Architectures

SNS + SQS Fan-out

One publish, N independent subscribers, fully decoupled.

5 sections · 3 gotchasread

Lambda + SQS

Serverless consumers with built-in scaling and partial-failure support.

5 sections · 3 gotchasread

Event-driven Systems

Why queues sit at the heart of every distributed system.

6 sections · 3 gotchasread