How to consume big parquet files in power bi ?

[jaryszek]

Hello,

How to consume big parquet files in power bi ? Where to put them within microsoft cloud architecture? 

Best,
Jacek

[jaryszek]

thanks. 

what is a difference between 

1. Connect through SQL views (Synapse) or direct lakehouse connection (Fabric) ??
   Best,
   Jacek

[jaineshp]

Hey @jaryszek,

You can find difference anywhere over the internet.

Here's a small part: -

Performance	Moderate (can have latency, especially serverless)	High (low-latency, optimized for Fabric)
Caching	Manual (materialized views, indexes)	Automatic OneLake caching
Data Modeling	SQL-based (views, joins, constraints)	Lakehouse tables, DAX or Power Query
Transformation	In SQL Views	In notebooks, pipelines, or Power BI

Fixed? ✓ Mark it • Share it • Help others!

Best Regards,
Jainesh Poojara | Power BI Developer

[jaryszek]

It is not a question how to find differences only. 

But when to choose specific solution? 

[jaineshp · Accepted solution]

Hey @jaryszek,

When to Choose SQL views (Synapse) and Direct lakehouse

1. Choose Direct Lakehouse Connection

When your parquet files are already clean and analysis-ready. Best for teams comfortable with Power Query transformations who need fast performance on large, well-structured data.

2. Choose SQL Views (Synapse)

When your parquet files need complex joins or transformations before consumption. Ideal for SQL-heavy teams who need to implement business logic at the database layer.

3. Decision Factor: Team Skills Over Technology

Choose based on your team's expertise - Direct Lake needs Power Query skills, SQL views need T-SQL knowledge. Work with your team's strengths, not against them.

Fixed? ✓ Mark it • Share it • Help others!

Best Regards,
Jainesh Poojara | Power BI Developer

[jaryszek]

the team is me with sql and power query expertise 🙂 

Thank you!!

[jaineshp]

Hey @jaryszek,

• Azure Data Lake Gen2 - Best storage option for parquet files
• Organize in folders by date/ID for better filtering performance

Option : Synapse Serverless SQL (Most practical)

• Create Synapse workspace
• Build SQL views over your parquet files
• Connect Power BI via DirectQuery
• Pay only for queries you run

Power BI Connection Steps

1. Never use Import mode for 70GB files
2. Always use DirectQuery connection
3. Connect through SQL views (Synapse) or direct lakehouse connection (Fabric)
4. Set up proper filters on ID column from the start

Fixed? ✓ Mark it • Share it • Help others!

Best Regards,
Jainesh Poojara | Power BI Developer