A company is migrating two simple Node.js applications from on premises to AWS. One app collects sensor data and writes to a MySQL database; the other aggregates data into reports. When aggregation jobs run, some load jobs fail. The migration must resolve the loading failures and occur without disruption or changes visible to customers. What should a solutions architect recommend?

The correct option addresses all requirements: AWS DMS ensures continuous, non-disruptive migration of the MySQL database. Using an Aurora Replica for aggregation jobs isolates read-heavy operations, preventing performance degradation on the primary database, which resolves the loading failures. Hosting collection endpoints as Lambda functions behind an ALB provides scalability and high availability for the application. Amazon RDS Proxy improves connection management for Lambda, which often creates many short-lived connections. Updating the DNS after synchronization ensures a seamless cutover without customer impact. The other options have drawbacks: Replicating directly to an Aurora MySQL instance without DMS might be more complex for continuous data synchronization during migration. Using an NLB for Lambda is less common than ALB, which offers more advanced routing and HTTP features. Hosting on EC2 instances behind an ALB is a valid approach but Lambda is generally more cost-effective and scalable for simple, event-driven applications like sensor data collection. It also doesn't explicitly mention isolating aggregation jobs to a replica. Using Kinesis Data Stream and Firehose introduces unnecessary complexity and a different data ingestion pattern than the existing direct database writes, potentially requiring application changes.