Forum Discussion
Power BI
I am fairly new to Power BI and I want to know how people are setting up their semantic models.
My company currently has a Power BI file connected to SQL that is uploaded to a workspace in Power BI (when uploaded it spits out a samantic model). Then, they point to that file to create any reports. If they want to add more DAX logic or make any changes to the relationships, we have to open the main semantic model file, make the changes, and re-upload the file.
Is this how a proper Power BI build looks like?
I tried to create a dataflow and assumed that since it was being created in Power BI Desktop, I would be able to use it directly like a semantic model and connect my reports. When I connected my Power BI file to the dataflow and published it in my workspace, it also created a semantic model.
I am not sure what a typical build in Power BI looks like. Where can I get some guidance? Can someone share how you build your data out? Am i suppose to have so many samantic models? What if I want to use the same buisness logic created by the DAX in different reprots. What is the best way to set this up?
2 Replies
Hi CyberQueen
First, dataflows are primarily used for data transformation and cleansing, not as semantic models. They prepare raw data for ingestion into Power BI datasets.
A typical workflow involves:
Creating dataflows in Power BI Service for preprocessing.
Connecting datasets in Power BI Desktop to the dataflows for further modeling and DAX logic.
About how yo build your data, each workspace or project should ideally have one semantic model that acts as a reusable data source. Creating multiple datasets with similar logic leads to inefficiency and inconsistency. You can publish a semantic model to a centralized workspace and allow other workspaces or teams to connect to it using live connections.
Connect to semantic models in Power BI - Power BI | Microsoft Learn
To ensure DAX business logic is reusable across reports, you should keep all DAX calculations in the central semantic model.Best Regards,
Jayleny
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.
Thank you for the reply.
A follow up question if I create a dataflows and a semantic model (A) using most of the dataflows.
Then create a new sematntic model and need data and DAX that was created in the first semantic model. Is is best practice to connect Semantic model A and all the dataflows in my new model or just to create it all from scratch with the dataflows and recreate the DAX?
