2 equal models with different performance
Hey, I am a bit confused right now with Power BI CPU consumption.
We have a model that brought our capacity to its limits. 100% of a P3 capacity with 2-3 queries.
A user caused a 1.1M CPUms query.
The model was since refreshed and doesn’t cause any more problems CPU wise.
Yet while analysing the issue, I put a copy of the model in a different workspace to test things out. The models are completely equal. Same model, same data, same capacity, just different workspace.
I used the same DAX query in DAX Studio with clearing cache.
The new model takes 3x the CPU and takes double the time in duration as the original one. Exactly the behaviour that happend with the origjnal model.
How is that possible? What am I not seeing?
Git tells me the models are equal, the Query Plan is the same, just the performance is crazily worse. Copilot gave me a hint about vertipaq segmentation, but can that really be the issue? Something I cannot really control?