Slow things usually just turn out slow when they reach production. The worse scenario arises when a specific combination (often involving lots of data) occurs, and it is hard to simulate it on stage environments. Once you rule out obvious mistakes, the best way out the sluggishness is to monitor how your application is performing through new versions. This way you can identify slow components and treat them accordingly. This article is about how you can store performance data about your REST application using CloudWatch and visualize it with Grafana.
Read