Forum Discussion
Kudo CommanderRecommendations to Improve Power BI Incremental Refresh Performance
Hi manoj_0911
For your setup (Premium Capacity, Import Mode, Incremental Refresh, SQL Server), there are no Premium capacity settings that significantly reduce SQL Server load during refresh. Most refresh performance gains come from query folding, partition design, source indexing, and refresh scheduling rather than capacity settings.
Recommended practices:
✅Enable Large Semantic Model Storage Format for better partition management and XMLA write performance.
✅Ensure full query folding for RangeStart/RangeEnd filters so only incremental partitions are queried.
✅Stagger refresh schedules if multiple datasets refresh concurrently to reduce SQL contention.
✅Monitor Capacity Metrics for CPU, memory pressure, and refresh concurrency bottlenecks.
Note that Query Caching improves report query performance only and does not accelerate dataset refreshes or reduce SQL load during Import refresh operations.
In practice, the biggest improvements usually come from optimized Incremental Refresh, proper SQL indexing, and refresh schedule distribution, not from Premium capacity settings.
Refer below 📌 Microsoft reference -
Large semantic models in Power BI Premium - Microsoft Fabric | Microsoft Learn
How to configure workloads in Power BI Premium - Microsoft Fabric | Microsoft Learn
💡 Helpful? Give a Kudos 👍 — keep the community growing
✅ Solved your issue? Mark as Solution ✔️ — help others find it faster
Best regards,
Rupasree Achari | BI & Fabric Analytics Engineer
Helper IHi,
A few recommendations that have worked well for improving Incremental Refresh performance are:
* Ensure the date column used for `RangeStart` and `RangeEnd` is a true Date/DateTime type.
* Verify that query folding is still working after all Power Query transformations, as this has a significant impact on refresh performance.
* Choose an appropriate refresh window so that only recently changed data is refreshed while historical partitions remain unchanged.
* Keep transformations as simple as possible before applying the incremental refresh filter.
* If you're using DirectQuery or Hybrid scenarios, review whether the storage mode and refresh policy match your workload.
Could you also share:
* Your data source (SQL Server, Fabric Warehouse, Synapse, etc.).
* Whether you're using Import, DirectQuery, or Direct Lake.
* The current refresh duration and dataset size.
These details will help the community provide more targeted recommendations for your specific scenario.