# Some Minor Metrics for Serverless Processing As part of the processing of a bunch of SQS messages ([[Speed of Pushing SQS Messages]]), I kept some metrics from processing requests with a Lambda. The upper chart shows the number of "Orders Received", which is effectively the number of requests (shown in orange). The other metric on the same chart is the "Orders Processed", which is the number of successfully processed requests. When everything is running smoothly the lines overlap exactly because we are successfully processing every incoming request. Watching the chart over time, we can see the effects of changing the Lambda concurrency setting. During normal times we are processing about 2.5k requests per minute. With a queue being loaded up it makes sense to increase our concurrency to process the queue faster. It's clearly visible how the number of concurrent executions affects the number of orders per minute that we process. We can choose how much we want to scale. If downstream systems are getting overwhelmed then we can lower the concurrency. If we are in a hurry to get orders loaded into a reporting system then we can increase the concurrency. When adjusting the concurrency, it's important to remember that the maximum concurrent executions for *all* Lambdas in a given account and region is 1,000 by default. You can request a quota increase from AWS. ![[Pasted image 20221209173837.png]] --- Created: 2022-12-09 08:39