Create a Dataflow vs Connecting to a published Semantic Model

[J_Castro]

What is the advantage(s) of a dataflow in the service compared to creating a semantic model/dataset in the Desktop, publishing it to the service and later creating multiple reports pointing to it?

I know that after connecting to a dataflow using the Desktop, PowerQuery can be used to make further data transformations. Whereas a connection to a live semantic model would only allow the creation of calc columns and measures (DAX).

On the other hand, a live semantic model would have the relationships and measures already in it.

[lbendlin · Accepted solution]

Remember that a dataflow is an intermediate entity. You cannot consume it directly.

Occam's Razor is a good guide when you need to decide between re-usability and storage size reduction, between de-coupling the developer from a slow data source  and the overall fastest processing etc.

[v-zhouwen-msft · Accepted solution]

Hi lbendlin ,

Hi @J_Castro ,

Regarding your question, data flow is not limited to Power BI; it can be created and used in other services, such as Power Platform (Power Apps).Using semantic models, you can reuse DAX calculations and relationships created for a model in other Power BI files.

Best Regards,
Wenbin Zhou

[lbendlin · Accepted solution]

Remember that a dataflow is an intermediate entity. You cannot consume it directly.

Occam's Razor is a good guide when you need to decide between re-usability and storage size reduction, between de-coupling the developer from a slow data source  and the overall fastest processing etc.