Amazon SQS
When to Use SQS
Recognise when a fully-managed cloud queue beats running your own broker.
SQS is the right choice when you are on AWS and want a zero-operations, infinitely-scaling queue for decoupling components, buffering bursts, and distributing tasks to workers — without running, patching, or clustering a broker yourself. It integrates natively with Lambda, ECS, SNS, and the rest of AWS, and you pay only for what you use.
SQS is hiring a fully-managed courier service instead of running your own fleet: you give it parcels and it scales and operates itself, but it offers a fixed menu of services — for bespoke routing or a permanent archive you need a different provider.
Key Concepts
1
It is intentionally simple: it lacks the rich routing of RabbitMQ (no topic/headers exchanges — you compose with SNS for fan-out) and the replay and stream-processing of Kafka (consumed messages are deleted, retention is at most 14 days). If you need complex routing, choose RabbitMQ; if you need a durable replayable event log or high-volume stream processing, choose Kafka (or Kinesis/MSK on AWS).
2
The sweet spot is straightforward producer-to-worker queueing and, with SNS, simple fan-out — with no infrastructure to manage.
3
What the interviewer is probing.
What the interviewer is probing.
4
1. "When is SQS the right choice?" *Probing:* the fit. *Stalls:* "On AWS." *Moves up:*
decoupling within AWS where you want a fully managed queue with no capacity planning and no brokers
to operate.
1. "When is SQS the right choice?"
5
2. "What does it not give you?" *Probing:* honesty. *Stalls:* "Nothing." *Moves up:* replay,
complex routing, message priority, and consumption by several independent consumer groups from one
queue.
2. "What does it not give you?"
6
3. "How does it compare with Kafka on cost?" *Probing:* the commercial angle. *Stalls:*
"Cheaper." *Moves up:* cheaper at low and moderate volume with no operational overhead; at sustained
very high throughput a managed Kafka can win.
3. "How does it compare with Kafka on cost?"
7
4. "What is the usual architecture around it?" *Probing:* the practical knowledge. *Stalls:*
"Just a queue." *Moves up:* SNS or EventBridge for fan-out into SQS queues, with Lambda or a
container service consuming and a DLQ per queue.
4. "What is the usual architecture around it?"