We can use avg_queue and be_session_rate to estimate if we are getting more traffic than what the backends can handle, to take corrective action such using a secondary cluster or using some lightweight URL endpoints. But, the problem is, these parameters can't be generalized and depends on the backend configuration ( For ex: Tomcat Worker threads, connection queue size), There can be a glitch due to which avg_queue size increased while everything else is normal such as traffic and backends. But, this can be generalized if we consider response time, usually increased traffic causes increase in response times ( Ex: request spends most time in tomcat connection queue). Average response times along with avg_queue allows better estimate of traffic surge. I have read through documentation, and couldn't find a internal state variable for average response time. I want to check if there is any internal state variable to achieve above, if not I am requesting a feature for average response time variable. Regards,Pratap
