You have received a complaint that during high load time the response time to clients increases several times. Nodes on which the application is running are not fully loaded. You need to identify the root cause and fix it.
Prometheus is already installed on monitoring namespace in the cluster.