In the mass of announcements made at Microsoft Ignite, I want to mention the SQL database in Fabric, which finally closes a gap we had in Power BI for many years.

In May 2022, Microsoft launched Power BI Datamarts in public preview. There was a lot of interest in that feature because, while moving the self-service BI experience to the cloud, we did not have the equivalent of “Access” for manipulating data. A professional developer knows how to activate an SQL Database on Azure. Nevertheless, the idea of having a relational database “included” in the Power BI license was appealing for many reasons (integrated security, simpler configuration, lower administration overhead), not just the licensing cost. Therefore, Power BI Datamart was very interesting because it could potentially solve the “Access” issue. However, there were too many limitations; no new features were released, and after more than two years, the Datamart feature is still in preview. Is someone using it?

However, with the SQL database, we now have (apparently) a full-feature Azure SQL Database without worrying about storage, provisioning, security, and so on. It just works.

Is this what we aimed for when Datamart was released? Definitely, yes.

Is it going to replace any of the other data stores in Fabric? It depends.

Microsoft provides a nice comparison between Lakehouse, Warehouse, Eventhouse, Fabric SQL database, and Power BI Datamart (yes, SQL database and Datamart – there should be a dedicated office somewhere in Redmond spending days to decide whether to use uppercase or lowercase in product names). You might try to spend time there to understand which one is good for your requirements.

However, you have a good reason to start playing with SQL database in Fabric if:

  • You have a few GB of structured data (SQL database can store up to 4 TB).
  • You are used to manipulating structured data in structured data using schema constraints to validate the content and make sure you reduce the garbage loaded in the analytical system.
  • You want complete control of the data types without compromises between original and destination data sources (for example, MONEY in a SQL Server data source does not have a matching data type in Fabric Warehouse).
  • You want to use SQL to implement transformations and write the result in a table (like MERGE to implement a slowly changing dimension).
  • You just want something that works today (and where Copilot could actually help you based on the enormous existing knowledge base about T-SQL).

Where is the bad news? Well, the entry point licensing is Fabric capacity F2 SKU. A Power BI user with a Premium per-user license cannot access this feature.

I would like a “Fabric per-user license” to help individual consultants play with it. Now, I completely understand that the existing consumption model makes more sense for resources that should be used for backend activities. However, a smaller version of a SQL Database, reduced in size and with other capabilities, could be the solution for many self-service Power BI users who do not have access to Azure resources and need a relational store that just works, reducing the world pollution caused by “SharePoint marts” and “OneDrive databases”.

Microsoft won a bet several years ago: they moved many business analysts from Excel to Power BI, changing (for good!) the careers of so many people. I would love to see another similar bet, lowering the budget required for more people to use SQL database in Fabric (or Power BI) for personal educational purposes.

Nevertheless, SQL database in Fabric is a very important and welcome piece of news.