Question

A retail company is building a new order-fulfillment workflow on AWS. Each order must reserve inventory, charge the customer's payment method, and schedule a shipment. Each step is implemented as a separate microservice with its own database, so a single ACID transaction across all three is impossible. If the payment charge fails after inventory has been reserved, the reservation must be released, and if shipment scheduling fails after payment succeeds, the payment must be refunded. The architects want a managed, serverless orchestration approach that provides visibility into each step, automatic retries, and reliable execution of compensating actions when a downstream step fails. Which design BEST meets these requirements?

Accepted Answer

Use AWS Step Functions to orchestrate the workflow as a saga, invoking each microservice via Lambda tasks and defining Catch/Retry states that trigger compensating tasks to undo prior steps on failure.

Answer

Publish an order event to an SNS topic that fans out to all three microservices simultaneously, relying on each service to detect and reconcile inconsistencies asynchronously.

Answer

Wrap all three microservice calls inside a single Lambda function using a try/catch block, and store an order status flag in DynamoDB to indicate partial completion.

Answer

Enable DynamoDB transactions (TransactWriteItems) across the three microservice tables to guarantee all-or-nothing atomic writes for every order.