I have a dataset and a couple of dataflows in a PowerBI Embedded workspace. The dataflows serve as sources for the dataset. The reports connected to this dataset are embedded on our internal employee portal.
Previously both dataset and dataflows were hosted in a PPU workspace, and all refreshes worked fine. However, the limited number of embed tokens ran out so I've had to switch it to an Embedded workspace.
Now my dataset no longer refreshes because it exceeds the memory limit (we have the A2 SKU with 5GB memory and 2 v-cores) and it takes the dataflows about 5 times as long to refresh. The dataset is only 281MB in storage size, so I'm struggling to understand how it takes more than 5GB to refresh.
To test I did temporarily up the memory limit to 25GB, which did allow refresh of the dataset. It is however not a feasible option to up the SKU permanently as it costs 10x more.
Can someone help me understand why my model is pulling more than 5GB of memory for a refresh? And why my dataflows take 5x longer to refresh in the embedded workspace than in PPU workspace?
Is there a way in which I can keep this data structure (dataflows and dataset running in PowerBI workspaces) and still embed my reports? Or do you have any suggestions for other options how to host my dataset (I've been looking at Azure data factory?)
Thanks in advance!
