undefinedHandling Large Source Tables with Many Columns – Model Size Limitation in Power BI

[AdithyaPythi]

Hi everyone,

I’m working with a source system that has a very large number of columns and tables. Due to Power BI model size and performance limitations, I cannot bring all columns into the data model.

Current challenges:

• High column count (wide tables)

• Large row volume

• Import mode hitting memory limits

• Performance degradation when loading full schema

I need most of the data for analytical flexibility, but including everything causes model size and refresh issues.

I’d appreciate guidance on:

1. Best practice for handling very wide source tables

2. Whether to:

   • Split fact tables?

   • Normalize in Power BI?

   • Use DirectQuery?

   • Move transformations upstream (SQL/Dataflow)?

3. Strategies for reducing model size without losing analytical capability

4. Experiences using Composite Models for this scenario

Would love to hear architectural recommendations from people who handled similar enterprise-scale datasets.

@PowerBI, @MicrosoftPBI