SAA-C03 - Design Resilient Architectures (26% of the exam) - Section 2.1

Design loosely coupled architectures using Amazon SQS, Amazon SNS and Amazon EventBridge to decouple components.

Describe how Amazon SQS queues buffer messages between producers and consumers to absorb bursts and decouple processing rates, and contrast this with Amazon SNS fan-out, which pushes a single message to multiple subscribers simultaneously. Use Amazon EventBridge rules to route events from AWS services and custom applications to targets without tight dependencies between the sender and receiver.

Amazon SQSAmazon SNSAmazon EventBridgeEvent-driven architecture

Practice question for this objective

Free sampleDesign Resilient Architecturesmedium

A company is choosing between Amazon SQS, Amazon SNS and Amazon EventBridge to ingest events from a third-party SaaS provider and route each event to different targets such as an AWS Lambda function or an AWS Step Functions workflow based on the event's content, without writing custom dispatch code. Which two capabilities are distinctive reasons to choose Amazon EventBridge for this design? Select TWO.

  • AEventBridge guarantees strict first-in-first-out ordering and exactly-once processing for every event delivered to its targets.
  • BEventBridge provides a poll-based queue where each consumer deletes messages after a visibility timeout to ensure single processing.
  • CEventBridge offers partner event sources that ingest events directly from supported SaaS providers onto an event bus without building a custom integration. Correct
  • DEventBridge rules match on event content and route matching events to targets such as Lambda or Step Functions, applying routing logic without custom dispatch code. Correct
  • EEventBridge stores messages for up to fourteen days so a single worker fleet can drain a backlog at its own pace during traffic spikes.
EventBridge is chosen for SaaS partner event ingestion and content-based routing to varied targets without custom dispatch code, unlike SQS buffering or SNS fan-out. EventBridge ingests events from supported SaaS providers through partner event sources and evaluates content-based rules to invoke targets such as Lambda or Step Functions declaratively, which together remove the custom integration and dispatch code that a plain SQS queue or SNS topic would require.

Why A is wrong: FIFO ordering and exactly-once delivery are properties of SQS FIFO queues, not EventBridge, which delivers events at least once without guaranteed ordering.

Why B is wrong: Visibility timeouts and poll-and-delete semantics describe SQS; EventBridge is push-based and routes events to targets rather than offering a consumer-polled queue.

Why C is correct: Partner event sources let supported SaaS providers deliver events straight onto an EventBridge bus, removing the need to build and maintain a bespoke ingestion path.

Why D is correct: Content-based rules evaluate the event payload and invoke chosen targets declaratively, which removes the custom routing code the team wants to avoid.

Why E is wrong: Durable backlog retention for self-paced workers is an SQS strength; EventBridge focuses on routing events to targets, not buffering a backlog for polling consumers.

See more SAA-C03 practice questions, answers explained.

Exam traps in Design Resilient Architectures

Answers that look right on this material and are not. Each one is a distractor from a different question in the SAA-C03 bank for this domain.

  • Amazon SQS, using one queue per target and a custom poller that reads each message, inspects its body, and forwards it to the correct downstream service.

    Why it is wrong: SQS only stores and delivers messages to pollers; the content-based routing to many target types would have to be built and maintained as custom dispatch code.

  • Increase the EventBridge rule retry policy maximum age and retry attempts so that the two-week-old events are automatically re-sent to the targets once the corrected consumers are live.

    Why it is wrong: Retry policy only governs delivery retries of recent failed invocations within a short maximum age; it cannot reach back two weeks or deliberately replay successfully delivered events.

  • Publish to one SNS topic per subscriber type and have the producer decide at publish time which topic each order event should go to based on its total and country.

    Why it is wrong: Splitting into per-subscriber topics pushes routing logic into the producer and multiplies topics, which the requirement explicitly rules out and increases coupling rather than letting SNS filter.

Examworthy is not affiliated with or endorsed by Amazon Web Services. Original, blueprint-aligned practice material only.