A couple of weeks ago, I was working on some AWS CDK-based code and I was trying to figure out how to configure auto-scaling for the provisioned concurrency configuration of an AWS Lambda function. We wanted to run some performance tests on our service and were wondering how scaling provisioned concurrency would impact our overall latency. We tried with the default configuration but also wanted to experiment with a bit more aggressive scaling policy so we had to provide our own metric configuration.
In this article, I will explain what provisioned concurrency is and how to set up an auto-scaling configuration for it using AWS CDK. We’ll be looking at using predefined metric configurations, but also how to do it with a custom metric.