A retail company runs multiple applications that register orders, process returns, and provide analytics. Currently, they use a MySQL transactional database and an Oracle OLAP analytics database, all on EC2. Several application components produce events from different sources; a weekly ETL copies application data to the analytics database. The company wants an event-driven, serverless redesign that provides near real‑time analytics. Which architecture meets these requirements?

The correct option provides a fully serverless, event-driven architecture. Refactoring into microservices on Amazon EKS with Fargate offers container orchestration without managing EC2 instances. Migrating to Aurora Serverless MySQL and Redshift Serverless provides scalable, managed databases that align with the serverless goal. Amazon EventBridge is ideal for routing events between microservices and the analytics system, enabling near real-time processing. Incorrect options: ECS with Fargate is good, but keeping MySQL on EC2 is not serverless. Amazon Neptune is a graph database, not suitable for general OLAP analytics. SQS is a queuing service, not an event router. Auto Scaling EC2 instances is not a serverless approach. Aurora MySQL is good, but SNS is a pub/sub messaging service, not an event router for diverse event sources to multiple targets. AppStream 2.0 is for streaming desktop applications, not for hosting backend microservices. AWS IoT Core is for IoT device data ingestion, not general application events.