When we were choosing new APM package, one of the statements was that DT agents would have no or very limited impact on the systems (performance,...) During stresstests of a new application, we did a test where on one of the clustered webservers DT-agent was started, on the other one DT-agent was not started. We observe that, tough the load is perferctly spread over the 2 webservers, cpu usage in the webserver that has DT-agent active is double in comparison with the webserver that has no DT agent active.
This observation makes our system engineers quite nervous, as if it looks like impact on system is not minimal anymore.
Does anyone has simular observations ? Is this normal behaviour ?
-Monique-
Answer by Monique V. ·
Kristof,
Thanks for your answer. Your perhaps is right on the spot, all is being captured what (as you say) can explain the overhead. We are planning to ask assistance of a DT consultant to guide us in making the right choises for filtering data (and for some other questions)
Monique
Ok. It looks like we might be in touch then (I am based in Belgium as well). You can reach me at kristof.renders@dynatrace.com if needed.
Answer by Kristof R. ·
Hi Monique,
Can you give some insights on how the instrumentation has been configured? Perhaps a lot (if not all) web request headers and parameters have been captured which could explain the overhead.
KR, Kristof
JANUARY 15, 3:00 PM GMT / 10:00 AM ET