The correct answer is Store the order in the database. Send a message that includes the order number to an Amazon Simple Queue Service (Amazon SQS) FIFO queue. Set the payment service to retrieve the message and process the order. Delete the message from the queue.

This solution uses a **FIFO (First-In, First-Out) SQS queue** to decouple order creation from payment processing. By including the order number as the **message deduplication ID**, SQS ensures that even if the message is sent multiple times (e.g., due to retries or timeouts), only one copy will be delivered within the deduplication window. This architecture preserves exactly-once order processing behavior, which prevents duplicate orders and aligns with distributed system best practices.

Configure the web application to send an order message to Amazon Kinesis Data Firehose. Set the payment service to retrieve the message from Kinesis Data Firehose and process the order is incorrect because Kinesis Data Firehose is primarily used for streaming data to storage destinations like S3 or Redshift, not for event-driven application messaging or transaction deduplication.

Create a rule in AWS CloudTrail to invoke an AWS Lambda function based on the logged application path request. Use Lambda to query the database, call the payment service, and pass in the order information is not a scalable or timely solution. CloudTrail is designed for auditing, not real-time application logic triggering, and has significant latency.

Store the order in the database. Send a message that includes the order number to Amazon Simple Notification Service (Amazon SNS). Set the payment service to poll Amazon SNS, retrieve the message, and process the order is incorrect because SNS is a pub/sub service without message retention or deduplication guarantees. It does not support polling or safe reprocessing in case of duplicate submissions.

Why this question matters: It tests your ability to use decoupled architecture patterns like queues to solve idempotency, retry handling, and system resilienceessential skills for designing fault-tolerant distributed systems in AWS.

Tip: Look for FIFO queues when you need **deduplication, ordering, and exactly-once processing**. Also, always decouple stages like order creation and payment to ensure reliability and fault isolation.

Full AWS Practitioner Certification Question