Explicit Measures Podcast 403: Discussing Composite Models In Power BI (Again)
Mar 4, 2025
Discover the evolving landscape of composite models in Power BI! The hosts explore the advantages and challenges of integrating PySpark within lake house environments. Learn about new features like Direct Lake and their impact on reporting tools. They discuss the necessity of clear governance when using composite models to ensure data integrity. Plus, insights on how Microsoft Fabric redefines data storage and management highlight the potential for better analytics. Tune in for a deep dive into effective data modeling strategies!
AI Snips
Chapters
Transcript
Episode notes
Initial Promise of Composite Models
- Composite models were initially exciting due to the ability to expand live connections to semantic models.
- This allowed adding data sources without altering the original model, which seemed revolutionary.
Unexpected Behavior of Composite Models
- Tommy Puglia recalls testing composite models and encountering unexpected results.
- Filtering behaved differently than with imported models, and DAX measures produced incorrect aggregations, revealing hidden complexities.
Performance Implications of Composite Models
- Be mindful of performance implications when using composite models, especially with direct query over Analysis Services.
- One visual can trigger multiple queries, significantly increasing processing time and causing slow performance.