estimating or monitoring APM overhead

Question

We are looking at expanding our use of APM to monitor servers at remote sites with low bandwidth connections.

Is there a way to estimate the overhead that APM will use for a set of operations? Or, failing that, to measure it? The only way I can think of would be to use NTA to compare total traffic from the poller to a server over a period of time, and compare utilization with and without APM enabled.

aLTeReGo · Accepted Answer

Leveraging data collected via NPM and NTA would be the way to get the most accurate measurement. I've done some rough approximations (read "spitballing") for common component types using default polling intervals below. These numbers are on a per-component monitor basis, so if you have multiple component monitors within a given application template you'll need to account for these accordingly. These are also averaged over time.

WMI Component Monitor = 2kbps
SNMP Component Monitor = 0.01Kbps
RPC 19Kbps

As you can see, when dealing with high latency, low bandwidth connections you'll want to leverage SNMP wherever possible. I recommend using the SNMP Service and SNMP Process component monitors in lieu of WMI or RPC. In the case of Windows Performance Counters, most every counter is available via WMI which uses less bandwidth than RPC.

estimating or monitoring APM overhead

Top Replies