Data transformation in PBI desktop is very slow when updating

[mkrcmar]

I am responsible for several datasets (semantic models). While working on any updates of data transformation, the most of my day is spent by waiting untill PBI finishes the background tasks and let me continue with another step. Is there any way to tell PBI to remeber whatever has been already loaded before and do not keep reloading and recalculating everything every time from the beginning? Any practical tips how to spend less time waiting on recalculations in data transformation? Thank you.

[lbendlin · Accepted solution]

No. Power BI has no memory. You can decouple the query refreshes with XMLA (where you can selectively refresh individual tables).  You can consider dataflows to get faster ingestions (and some more refresh decoupling)

A default semantic model refresh will always include a complete recalculation of all calculated columns or tables.  Keep these to a minimum and make the formulas fast.

For very large amounts of data (more than 2.5 hrs of refresh) consider using incremental refresh.