Indeed if I were to configure the load-balancer I would limit both the number of concurrent requests and the number of queued requests. (In fact this is what I

Indeed if I were to configure the load-balancer I would limit both the number of concurrent requests and the number of queued requests. (In fact this is what I do in my HAProxy deployments.) However I assume that the customer (the original author was referring to) didn't want to apply any limitations. (Why is beyond my reasoning...)