How to find which interactive operations account for the CPU consumption?

[Quiny_Harl]

Hi folks,

I have a Power BI report that generates significant CPU consumption and it accounts for a big portion of the whole premium capacity consumption. As a result, the capacity becomes throttled. What we know is most of this consumption comes from interactive (query) operations. For this reason, we want to optimize the report.
However, it is a huge report that contains many pages, measures and visuals. It is difficult to assess which interactions specifically account for the majority of the CPU consumption. I already know which visuals have the highest loading time, but how do I know, for exmaple, if these visuals actually account for the high consumption? I can also have the number of views per page with the usage metrics report which could help, but I am wondering if a better approach/tool exsists.
Is there a way to find out which specific query operations generate how much of the CPU consumption? Can Capacity Metrics App, for example, provide this info?
Any ideas/solutions are welcome!

[lbendlin]

The obvious but very expensive answer is to use Azure Log Analytics on the workspace that you suspect of hosting the offending queries.

Capacity Metrics doesn't go down to query level.

Try to isolate the user causing the CUs and watch them use the report.

[Quiny_Harl]

@lbendlin, thanks for your input.
Do you mind explaining why the Azure Log Analytics solution is so expensive and how would that work?

[lbendlin]

Azure Monitor Logs cost calculations and options - Azure Monitor | Microsoft Learn