Auto-updated semantic model (dataset) in Fabric workspace

[lyjavier]

Hi,

I am currently facing an issue with the semantic model for a multi-tenancy case in the Fabric workspace.

The scenario is as follows:

• I have a report with a direct query connection mode (to the lakehouse of tenant A) in Power BI Desktop. After publishing it to the workspace, I have one report (let’s call it "report") and one semantic model (let’s call it "tenantA"). There are many other tenants, such as tenant B, tenant C, tenant D, etc., each with its own lakehouse. Whenever I make changes to the report that require updates to the semantic model for tenant A, I have to download the report, open it in Power BI Desktop, change its SQL endpoint to the lakehouses of tenant B, tenant C, tenant D, and then republish them to the workspace to replace their old semantic models. This is a common scenario known as multi-tenancy. The report is embedded in our web app, allowing tenants to see their dashboards with the corresponding semantic models that refer to their lakehouses.

Is there a way to automate this process at the moment?

Thank you in advance!

[lyjavier]

Not aware of this, how can it work for this scenario? Can you walk through me a little bit of this?

[SachinNandanwar]

Honestly, I am unsure if the library could cover your scenario.You could check more details below.

https://github.com/m-kovalsky/Fabric?tab=readme-ov-file#dataset-refresh

https://learn.microsoft.com/en-us/fabric/data-science/semantic-link-overview

[lyjavier]

Hi, I've been working with sempy, it's looking good to create a semantic model from bim model file (belong to an existing semantic model), just update the connection string the lakehouse name. But there is new issue with the refresh that semantic model, in Power BI Desktop, I have created a calculated column within a table, but the sempy_labs.refresh_dataset seems not working for semantic model with direct query connection mode? So can I have another approach to make it automatically?

[SachinNandanwar]

Have you tried the SemPy library for semantic model ?