Comparing query design efficiencies in Power BI, with text files, lakehouse & database tables
I wanted to share the results of a few experiments I recent conducted with one of my favorite sets of sample data. This will take at least two blog posts to cover, but I will summarize them here:
Compare data load & transformations with CSV files vs a Fabric lakehouse using the SQL Server connector:
Loading 20 million fact rows from CSV files vs a Fabric lakehouse, using Power Query.
Same comparison with deployed Power BI model.
Comparing Fabric data transformation options & performance:
Loading & transforming the same data with Power BI Desktop, Fabric Gen2 dataflows, Fabric pipelines and Spark notebooks.
Comparing semantic model performance in Fabric and Power BI:
Report & semantic model performance comparing the same data in Import mode, DirectQuery and Direct Lake.