Serverless Tip #6: Pick the optimal memory configuration for your functions

The only configuration option for performance on a Lambda function is the memory slider, which goes from 128MB to 10GB.

Lambda functions get CPU credits proportional to the amount of memory. In many cases, going from 256MB to 1024MB will give you a 4x performance boost without increasing cost.

Going to 10GB might not result in a further 10x performance boost since your code might not utilize the extra CPU cores available. In that case, you are only increasing costs and not performance.

The Lambda Power Tuning tool can analyze your functions and suggest the best configuration to minimize cost and/or maximize performance.