For any performance enhancement that's not trivial to implement and obviously advantageous, we need a way to quantify the impact. We'll want both debug logging within the functions to be able to see what aspects of the execution are taking significant time and tooling to capture end-to-end request timing, since many of the potentially time-consuming pieces (e.g. cold starts) happen outside the actual Lambda function.
For any performance enhancement that's not trivial to implement and obviously advantageous, we need a way to quantify the impact. We'll want both debug logging within the functions to be able to see what aspects of the execution are taking significant time and tooling to capture end-to-end request timing, since many of the potentially time-consuming pieces (e.g. cold starts) happen outside the actual Lambda function.