Hello Zdenek,

Thank you for your reply! There are a number of reasons for high memory usage during refresh. Due to the transactional nature of Power BI’s data refresh, the entire refreshed model is loaded into memory with the old model, then after refresh is complete the old model is purged, meaning you should expect at least twice the memory usage during refresh, and perhaps more depending on the optimization of your M queries. If memory usage remains high after refresh, I would agree this is likely Power BI still compressing the data.

I would recommend running Query diagnostics,
https://docs.microsoft.com/en-us/power-query/querydiagnostics , and depending on your model needs implementing incremental refresh to reduce memory usage on refresh, https://docs.microsoft.com/en-us/power-bi/connect-data/incremental-refresh-overview.

Best regards,
-Clayton

thank you for a great article. I’m facing a strange behavior I don’t understand with my PBI model. When I check the model size with DAX Studio in PBI Service before and after refresh of the dataset after the refresh the model size is x times bigger (for some datasets 5 times, for other 3 times..) than before refresh. After some time the size goes automatically back to the size before the refresh. Does it mean that after the refresh the model in PBI Service is uncompressed and after some time the compression brings the size back to the compressed before refresh state?
Thank you very much for your response.
Best regards,
Zdenek Soldan