Monitoring a rapidly oscillating Performance Monitor counter

mitchross

I'm having challenges monitoring a application with SolarWinds.  We have an application (OSI PI) that processes data. I want to make sure data is regularly moving through the system - and alert me if a counter is showing low/no traffic. Sounds easy enough, right?

The way this app works,  is that is sends a burst of data, and is idle for a second or two, then another burst.  (See screen shot below)

We've got SolarWinds polling at various intervals, and the problem is that it often polls that Perf Mon counter between bursts of data, and sees a 0.

When monitoring 24x7, odds are that eventually the poller sees 5 or 10 of these 0's in a row, and sends off an alert that the system is down.

Anyone have a suggestion on how to monitor this?

rharland2012

What are sure indicators that this application has failed? Do ports stop listening? Do processes/services stop or become unavailable?

If you can define a system state that shows accurately 'the app is no longer functional', I'd worry less about data transfer and more about monitoring that system state.