Facts and Dims

[dpjatochh]

Hi everyone,

How does everyone approach modeling in Dataflows Gen2? For our client it seems easier to do it with a dataflow because they are not on a sufficient SQL or notebook level.

Would you create a dataflow for every fact/dim? It makes sense for me to keep it seperate. But i'm also thinking - if multiple facts/dims use the same source tables that we might combine them into one dataflow

Thanks

[Pragati11 · Accepted solution]

HI @dpjatochh ,

I would suggest here to go by Dataflow best practices.

• You can have first dataflow to bring in your raw data. (Staging Dataflow)
• You can have second dataflow to do all the transformations like creating DIM and FACT tables. (Transformation Dataflow)
• Finally you can have a dataflow having all your final transformed tables (DIMS and FACTS); which you can use further to create your semantic model. (Output Dataflow)

Staging DF --> Transformation DF --> Output DF

You can read about dataflow best practices here:

https://learn.microsoft.com/en-us/power-query/dataflows/best-practices-developing-complex-dataflows