Managing and Optimizing Semantic Model Refreshes

Data Ingestion and Semantic Models with Microsoft Fabric

Alex Kuntz

Head of Cloud Curriculum, DataCamp

Data Refresh in Semantic Models

What is Data Refresh?

  • Ensures latest data is visible in reports

Refresh Requirements for Storage Modes:

  • Import Mode: Requires scheduled refresh for updates
  • DirectQuery Mode: No refresh needed; queries live data
  • Direct Lake Mode: No refresh needed; accesses OneLake directly

Data Refresh in Semantic Model

Data Ingestion and Semantic Models with Microsoft Fabric

Scheduled and On-Demand Refresh

  • Scheduled Refresh

    • Automates refresh at specific intervals
    • Ideal for regular updates without manual effort
  • On-Demand Refresh

    • Manually triggers refresh anytime
    • Flexibility for immediate updates
    • Useful for real-time scenarios

Semantic Model Refresh Options

Data Ingestion and Semantic Models with Microsoft Fabric

Incremental and Full Refresh

Incremental Refresh

  • Refreshes only new or modified data
  • Reduces time and resource usage
  • Ideal for large datasets with regular updates

Full Refresh

  • Reloads the entire dataset
  • Ensures complete data consistency
  • Suited for smaller datasets or full data replacement scenarios

Data Refresh in Semantic Model - Incremental

Data Refresh in Semantic Model - Incremental

Data Ingestion and Semantic Models with Microsoft Fabric

Semantic model refresh activity

  • Automates dataset refreshes to keep Power BI reports up-to-date.
  • Integrated within Microsoft Fabric's Data Factory.

Semantic Model Refresh Activity

Key Features:

  • Connection: Requires a pre-configured connection to Power BI datasets.
  • Default Behavior: Runs a full data refresh by default, replacing all data.
  • Preview Feature: Currently available in preview, enabling early access to try it out
Data Ingestion and Semantic Models with Microsoft Fabric

Let's practice!

Data Ingestion and Semantic Models with Microsoft Fabric

Preparing Video For Download...