Which solution will meet these requirements with the LEAST amount of change to the application?

Explanations:

Updating the Lambda functions to open the database connection outside of the handler could improve connection reuse, but it may not significantly reduce latency under high load. Increasing provisioned concurrency alone might not address database connection limits during traffic bursts.

Creating an RDS Proxy allows for connection pooling and efficient management of database connections, reducing latency and increasing resilience during traffic spikes. Storing database secrets in Secrets Manager enhances security, and updating the Lambda functions to connect through the RDS Proxy minimizes code changes while improving performance. Increasing provisioned concurrency also helps manage bursts in traffic.

While increasing the max_connections parameter can allow more concurrent connections to the RDS instance, this option does not address connection management or pooling, which can lead to connection issues during traffic spikes. Additionally, increasing reserved concurrency might not effectively handle sudden traffic increases compared to using RDS Proxy.

Although creating an RDS Proxy and using Secrets Manager is a good solution, increasing reserved concurrency instead of provisioned concurrency does not effectively prepare the Lambda functions for sudden bursts in traffic. Provisioned concurrency allows for better handling of spiky workloads, while reserved concurrency limits the maximum capacity, which is counterproductive in this context.