I have a Microservices-based ordering application that is deployed on a Kubernetes cluster, we are observing high response time for the Initial request in the below two cases. 1. If we do a rolling bounce there is a high response time for the initial request? 2. If the system is idle for a couple of hours we observe a high response time for the requests. (The ordering application comprises of 40 + different Microservices.)

Is there any configuration we can do to avoid this behavior for the initial request? Or we should have some kind of warm-up script to address this problem.

Would you please advise how you have handle such cases?