Reusable semantic model Best Approach

[Koritala]

Hi,

I just want to know that what is the best approach to build the semantic model to avoid the duplication of creating tables.

Which is the best among below 2 appraoches.

1. create dataflows and call these into semantic model as source using get data option

2. create a master semantic model in desktop with the common tables and publish to service and use these tables in child semantic model using get data and semantic model option.

Can you please explain the pros and cons between above 2 approaches?

My goal is avoid creation of same table multiple times so that we can restrict the multiple query submmisions for same table  to backend when multiple semantic models are scheduled for refresh.

Wouls appriciate if anyone can cleary explain the best approach.

Thanks,

Sri

[rohit1991 · Accepted solution]

Hii @Koritala 

For avoiding duplicate table creation and reducing backend query load, using Dataflows as the source for multiple semantic models is the best approach. Dataflows extract and refresh data once, store it centrally, and then all semantic models reuse the same prepared tables, so the source system is not queried repeatedly. Creating a master semantic model and using it in child models helps with reuse, but it does not fully prevent multiple query executions during refresh and creates tight coupling. Best practice is Dataflows for shared data + separate semantic models for modelling and measures.