We are just rolling out an installation of APM across our WAN and it seems that collecting performance counters is very expensive in terms of bandwidth. Running some tests it seems that adding the colelction of one performance counter costs 450bps (approx). Given that we are monitoring 5 applications with 4 monitors each that turns into roughly 9000bps (for only 20 numbers). Of course this only happens every 5 minutes so it is pretty taxing on the WAN during polling times (we have traffic shaping where the control system gets priority). Can APM stagger this communication? It does seem like our traffic shaping is killing some of it as we sometimes have polls time out.
Maybe there is a better way to do the monitoring we want. I want to see if there is a memory leak or if there is a change in the PID of the application (it restarts on its own and we want to track that). The only I've found to get access to the PID is to grab the performance counter of it.
Any ideas would be helpful.
Tim