Forum Discussion

powerbiexpert22's avatar
powerbiexpert22
Icon for Impactful Individual rankImpactful Individual
2 days ago

incremental refresh

the semantic model size is around 3 GB ( in pbix) and takes lot of time to download, currently all tables are in import mode . i am planning to implement incremental refresh however the data from sou...
Need Help
power bi
sandeephijam's avatar
sandeephijam
Icon for Helper II rankHelper II
2 days ago

Hi powerbiexpert22​ 

Adding to the above krishnakanth240​ suggestions :

I'd prioritize:

Phase 1

  1. Find the largest fact tables contributing to the 3 GB model.
  2. Confirm which tables actually need Incremental Refresh.
  3. Determine the maximum historical correction/deletion period.
  4. Apply Incremental Refresh only to appropriate large fact tables.
  5. Keep smaller dimension/reference tables on normal refresh.

Phase 2, only if necessary

If historical changes have no time limit, introduce:

Source → SQL/Lakehouse → Power BI

and implement proper change/deletion handling there.

Phase 3, advanced

Use PPU + XMLA when you need targeted historical partition management rather than continually widening the normal refresh window. XMLA provides advanced partition-management capabilities.

Bottom line

Don't build CDC first. First determine the historical change/deletion window.

If the business says," Anything older than 12 months never changes," your problem becomes much easier. Incremental Refresh with a 12-month refresh window may be all you need.

If the business says," We can delete a five-year-old transaction tomorrow," then I would move to the staging + change tracking + controlled Power BI partition refresh architecture.