Blogs

Understanding Refresh Options in Custom Semantic Models in Microsoft Fabric

Inkey Solutions, May 20, 2025204 Views

When developing Power BI reports using Direct Lake mode in Microsoft Fabric, it is critical to understand how data refresh operations work within semantic models.

Direct Lake offers a new approach by enabling models to query data directly from One Lake, eliminating the need for traditional import processes and delivering real-time analytics with high performance.

In this post, we’ll cover the two key refresh options available for custom semantic models — “Keep your Direct Lake data up to date” and “Configure a refresh schedule” — explore different use cases where each option becomes critical and explain how schema and data changes are handled based on configuration settings.

Refresh Options available on Power BI Service for Direct Lake storage mode

• Keep your Direct Lake data up to date
  • This option controls how Power BI handles real-time data updates from One Lake for semantic models using Direct Lake storage mode.
  • This option can be found under the Refresh section of the semantic model settings.
  • If this option is
    • Enabled:
      Power BI automatically detects and reflects any data changes made in One Lake directly in the semantic model and corresponding Power BI reports. No manual intervention or scheduled refresh is needed to see the latest data.
    • Disabled:
      Data changes in One Lake are not immediately reflected. Updates will only appear after either a manual refresh or a scheduled refresh operation is performed.
• Configure a Refresh Schedule
  • When a refresh schedule is configured, Power BI updates the semantic model according to the defined schedule, pulling in the latest data from One Lake or other connected sources.
  • Scheduled refresh becomes especially critical when the “Keep your Direct Lake data up to date” option is turned OFF, ensuring that data remains current without requiring manual refresh actions.
  • When the “Keep your Direct Lake data up to date” option is turned ON, configuring a schedule refresh has no impact. Since Power BI is already automatically detecting and reflecting data changes in real time from One Lake, the scheduled refresh becomes largely redundant for maintaining data freshness.

[Fig. 1 Direct Lake storage mode]

Refresh Options available on Power BI Service for storage modes other than Direct Lake

• Keep your Direct Lake data up to date
  • This option is no longer visible under the Refresh section of the semantic model.
  • This change happens because the semantic model has no table with Direct Lake storage mode.
• Configure Refresh Schedule
  • A refresh schedule was defined to import data from the source into the semantic model at specified intervals. This ensures that data is regularly updated without manual intervention.
    

[Fig. 2 Storage mode other than Direct Lake]

Important Note
Regardless of whether “Keep your Direct Lake data up to date” setting is enabled or disabled, schema changes—such as adding or removing columns in the underlying Lakehouse tables—are always automatically reflected in the semantic model and Power BI reports. This ensures structural consistency without requiring additional refresh actions.
Additionally, note that the Power BI report may crash if the columns or tables used in the report are deleted from the underlying semantic model.

Conclusion

• Automatic Schema Change Detection:
  • Schema changes, such as adding or removing columns in Lakehouse tables, are automatically reflected in the Semantic Models and Power BI reports without requiring any refresh actions, regardless of the “Keep your Direct Lake data up to date” setting and manual or schedule refresh.
  • Because of this, the Power BI report may crash if the columns or tables used in the report are deleted from the underlying semantic model.
• Direct Lake Storage Mode:
  • With the “Keep your Direct Lake data up to date” setting enabled, data updates in Lakehouse are automatically reflected in Power BI, eliminating the need for scheduled refreshes. In this case, configuring the schedule refresh has no impact.
  • If disabled, manual or scheduled refreshes are required to update the data.
• Storage mode other than Direct Lake:
  • The “Keep your Direct Lake data up to date” option is no longer visible under the Refresh section of the semantic model as the semantic model has no table with Direct Lake storage mode.
  • The “Configure Refresh Schedule” option is available to import the data at a specific time.