Get certified for free when you join Fabric Data Days 2026 and dive into Fabric, Power BI, SQL, AI, and other essential data skills.
Join nowData Days is here! Join us now for 60+ days of learning, challenges, and connection. Learn more
I want to confirm my understanding of how DirectLake mode impacts Fabric capacity resources.
Question : Am I right to assume, if you are using DirectLake query mode, no "Background" CU is consumed? Rather only "Interactive CU" is consumed ( when users interact with the reports)? Or is some background CU used to keep the Delta Lake files / Parquet files up to date- for example when using database mirroring?
I can't see any specific mentions of Mirroring here : https://learn.microsoft.com/en-us/fabric/enterprise/fabric-operations
Any references or links would be appreciated.
Solved! Go to Solution.
Hi @Anonymous ,
Yes as you said, Interactive CUs are mainly used during user interaction with reports. When a user queries or interacts with a report powered by the DirectLake schema, the system uses the interactive CUs to process these requests in real time.
Background CUs may not be mentioned directly in the context of DirectLake operations, but it is important to understand that background CUs are typically used by operations that run in the background without immediate user interaction. This includes data refresh operations, application model changes, or other maintenance tasks.
DirectLake mode primarily affects interactive CUs because it enhances the user interaction experience by providing faster data retrieval and analysis capabilities. However, keeping incremental lake files or parquet files up-to-date (e.g., through database mirroring or scheduled refreshes) can consume background CUs, depending on the specific operations required to keep data up-to-date.
While the DirectLake schema itself focuses on optimising interactive analysis of large datasets, any background operations required to maintain data readiness (such as updating incremental tables or managing parquet files) will still follow the general principles of CU usage in the fabric.
Unfortunately, there is no direct reference to database mirroring in the DirectLake context in the provided documentation. However, it is reasonable to assume that operations related to data maintenance or updates running in the background will use the background CUs in accordance with the standard operations for fabric resources.
For more insight into DirectLake the following documentation may be helpful:
Learn about Direct Lake in Power BI and Microsoft Fabric - Power BI | Microsoft Learn
Learn how to analyze query processing for Direct Lake datasets - Power BI | Microsoft Learn
Best Regards,
Ada Wang
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.
Thanks for the detailed response.
Hi @Anonymous ,
Yes as you said, Interactive CUs are mainly used during user interaction with reports. When a user queries or interacts with a report powered by the DirectLake schema, the system uses the interactive CUs to process these requests in real time.
Background CUs may not be mentioned directly in the context of DirectLake operations, but it is important to understand that background CUs are typically used by operations that run in the background without immediate user interaction. This includes data refresh operations, application model changes, or other maintenance tasks.
DirectLake mode primarily affects interactive CUs because it enhances the user interaction experience by providing faster data retrieval and analysis capabilities. However, keeping incremental lake files or parquet files up-to-date (e.g., through database mirroring or scheduled refreshes) can consume background CUs, depending on the specific operations required to keep data up-to-date.
While the DirectLake schema itself focuses on optimising interactive analysis of large datasets, any background operations required to maintain data readiness (such as updating incremental tables or managing parquet files) will still follow the general principles of CU usage in the fabric.
Unfortunately, there is no direct reference to database mirroring in the DirectLake context in the provided documentation. However, it is reasonable to assume that operations related to data maintenance or updates running in the background will use the background CUs in accordance with the standard operations for fabric resources.
For more insight into DirectLake the following documentation may be helpful:
Learn about Direct Lake in Power BI and Microsoft Fabric - Power BI | Microsoft Learn
Learn how to analyze query processing for Direct Lake datasets - Power BI | Microsoft Learn
Best Regards,
Ada Wang
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.
Don't miss out on Data Days, June 15 through August 7. Learn Fabric, Power BI, SQL, AI and more.
Check out the May 2026 Power BI update to learn about new features.
| User | Count |
|---|---|
| 16 | |
| 10 | |
| 10 | |
| 8 | |
| 7 |
| User | Count |
|---|---|
| 40 | |
| 36 | |
| 34 | |
| 28 | |
| 20 |