Dynatrace - Method hotspots Tools - What is the unit ?

[billy_thach]

Hello !

May be someone could help me : i don't understand what is the unit used with Method Hotspots ? I want use this tools to understand the cpu consummation as indicated but i only see contribrution and i don't found this information in documentation ..

I join a screenshot to show metrics which i need unit information.

Thanks!

[skrystosik]

This number represents samples count. During all transactions dynatrace is sampling cpu. Amount of samples is also representation of time spend on those methods because samples are making in regular timeframes.

[billy_thach]

Thanks Sebastian for your answer!

So sum up, if i understand your explaination : this number represent the time spent on CPU ?

[skrystosik]

Not directly, It doesn't mean number of ms spent on CPU. It means that when CPU is sampled in 10ms steps, this amount of samples has been recorded.

[vladislav_samoy]

If I want to see Method Hotspots for a single purepath, does it mean Count = Method execution time or such methods exists in specified CPU samples?

[skrystosik]

It means that this method exists in X number of samples. It also means that method with higher number of samples was present on CPU longer. But you have to understand. CPU sampling is making in 10ms time periods. This means that short execution method can be missed. This also means that dynatrace doesn't know exactly how long execution taken (because we don't know if first sample was in exact place of starting execution time etc).

[billy_thach]

Thanks for your answer Sebastian : this metric is more clear for me. The CPU sampling with 10ms time is documentated somewhere ?

[skrystosik]

Actually this is information I use based on Appmon history. There CPU sampling from autosensor was fired in 10 ms periods. Actually I don't see documentation for method hotspots anyware in dynatrace documentation.

[skrystosik · Accepted solution]

Ok my bad I've made mistake. When you are anylizng single pure path there you will see number which is representation of contribution of response time. So it's not samples from cpu sampling. I think they are representing numerical representation of percentage of response time contribution.

Samples I was talking about are when you will go to whole service, and from this point of view you will try to analyze cpu consumption. Sorry for mistake 🙂

[billy_thach]

Thanks Sebastian! I have opened a ticket to the support, may be they can help us to clarify this point.

[vladislav_samoy]

Please post the feedback. I'm interesting in this topic as well