Best Practices in Fabric

[ahmedshalabyy12]

Dears,

Currently, I am working as a solo Data Engineer in Microsoft Fabric for the first time.

I have the following question:

I plan to use a Pipeline to ingest data from SharePoint, then clean, transform, and consolidate the data into a single table.

What is the recommended best practice for implementing a Medallion Architecture in this scenario?

Should I create three separate Lakehouses:

• Bronze Lakehouse for raw data
• Silver Lakehouse for cleaned and standardized data
• Gold Lakehouse for aggregated and business-ready data

Or would it be better to use a single Lakehouse and implement the Bronze, Silver, and Gold layers through separate notebooks (one notebook for each layer)?

Additionally, for the Gold layer, would you recommend using a Lakehouse or a Data Warehouse as the final serving layer? What are the advantages and disadvantages of each approach, and which option is considered a best practice in Microsoft Fabric for reporting and analytics workloads?
Thank you 

[pallavi_r · Accepted solution]

Hi @ahmedshalabyy12 

Would suggest towards a single Lakehouse with separate notebooks for Bronze, Silver, and Gold rather than creating three separate Lakehouses.
All three layers (Bronze/Silver/Gold) as tables/folders within the same Lakehouse
Using multiple Lakehouses can add unnecessary complexity, increase data movement, and make governance and monitoring harder. With a single Lakehouse, all layers can be managed as separate tables or folders, while notebooks handle the transformations from Bronze → Silver → Gold. This keeps the architecture simpler and aligns well with the Fabric Medallion approach.

For the Gold layer, as source is sharepoint, I would recommend to keep it simpler with a Lakehouse if you're primarily using notebooks, Spark, and Power BI (especially Direct Lake). Since it is sharepoint that needs to be put into consolidated table, Datawarehouse would be unneccessary. A Data Warehouse becomes a better choice when there is a strong SQL/T-SQL focus, a need for multi-table transactions, or advanced security features such as column-level security and dynamic data masking.

SharePoint → Bronze tables → Silver tables → Gold tables (within one Lakehouse) → Power BI

This keeps the solution simple, scalable, and easier to manage as a solo data engineer.

If this post helps, please accept this as a solution. Appreciate your kudos.

Thanks,

Pallavi