Which solution will resolve this issue with the LEAST operational overhead?

Explanations:

Adjusting the size of the Aurora MySQL nodes may increase connection limits, but it requires operational overhead for resizing and management. Adding retry logic in the Lambda functions could improve resilience, but it does not address the underlying issue of connection management and could lead to increased retries without resolving the connection timeout problem.

Implementing Amazon ElastiCache for Redis can help reduce load on the database by caching commonly read items, but it introduces additional complexity in managing a caching layer and may not directly resolve connection timeout issues related to Lambda’s concurrent connections.

Adding an Aurora Replica can distribute read queries, but it primarily benefits read scalability rather than addressing connection management issues directly related to Lambda’s concurrent invocations. This also introduces operational overhead in maintaining replicas.

Using Amazon RDS Proxy is the best solution as it manages connection pooling efficiently, allowing Lambda functions to share connections and reduce the number of new connections established. This minimizes connection timeouts with low operational overhead as RDS Proxy handles the complexity of connection management automatically.