In this related recipe we focus on the service side of the equation. Cloud-native databases, such as DynamoDB, respond in the low 10s of milliseconds, and the overall latency across AWS API Gateway and AWS Lambda for a BFF service should typically execute in the low 100s of milliseconds. So long as the database capacity is set appropriately and throttling is minimized, it would be hard to make a noticeable improvement in this performance from the end user's perspective. The only way to really improve on this is to not have to make a request at all.
This is a case where cloud-native can really be counter-intuitive. Traditionally, to improve performance, we would need to increase the amount of infrastructure and add an expensive ...