cancel
Showing results for 
Search instead for 
Did you mean: 
Reply
Dl_BL_PB Helper II
Helper II

Best Practise question for Modelling in Powerbi/Tabular - Avoiding Chained/Nested Chasm Traps

I am currently trying to understand the best practises around Powerbi and Modeling in Powerbi/Tabular (With big data). Below is an example of a snowflake design. (Client  which filters matter which then filters the facts). I just wanted to make sure i am following best practises. 

 

SimpleExampleSnowflake.png

My approach to a re-design is to split Client off Matter and have it directly pointing to the fact which will create a star schema Model. Is this approach correct? in saying that, the result of me creating a star schema will cause the Client to stop filtering items on Matter unless i created Bidirectional filtering from the fact to both client and matter. What would your approach be? I do not know the implications of having a huge model with bidirectional filtering which supports the star schema design.

4 REPLIES 4
Microsoft v-gizhi-msft
Microsoft

Re: Best Practise question for Modelling in Powerbi/Tabular - Avoiding Chained/Nested Chasm Traps

Hi,

 

Please try to make 'Dim Matter' to 'Fact Debt' and 'Fact Sales' reltionships to inactive and create relationships between 'Dim Client' and other two fact tables:

101.PNG

 

Hope this helps.

 

Best Regards,

Giotto Zhi

Ross73312 Community Champion
Community Champion

Re: Best Practise question for Modelling in Powerbi/Tabular - Avoiding Chained/Nested Chasm Traps

Another recommended approach is to merge your Client and Matter table into one single denormalised table.  Since both of your tables need to link via the Matter ID, this relationship will still work and you can still organise filters by the Client data.


   

              Did I answer your question? Mark my post as a solution!
       

Proud to be a Datanaut!


   


Highlighted
Dl_BL_PB Helper II
Helper II

Re: Best Practise question for Modelling in Powerbi/Tabular - Avoiding Chained/Nested Chasm Traps

I was considering that idea, however my matter and client dimensions are pretty big. So for end users i think i need to keep them split up. 

Ross73312 Community Champion
Community Champion

Re: Best Practise question for Modelling in Powerbi/Tabular - Avoiding Chained/Nested Chasm Traps

Being big is actually the reason to merge them.  The way the engine stores and compresses the information will likely mean it won't make a large change to the model size, but there is likely to be a performance gain.

 

From an optimisation standpoint, you can remove unnecessary columns, although from your screenshot you seem fairly lean already, which is a good thing.


   

              Did I answer your question? Mark my post as a solution!
       

Proud to be a Datanaut!


   


Helpful resources

Announcements
New Ranks Launched March 24th!

New Ranks Launched March 24th!

The time has come: We are finally able to share more details on the brand-new ranks coming to the Power BI Community!

‘Better Together’ Contest Finalists Announced!

‘Better Together’ Contest Finalists Announced!

Congrats to the finalists of our ‘Better Together’-themed T-shirt design contest! Click for the top entries.

Arun 'Triple A' Event Video, Q&A, and Slides

Arun 'Triple A' Event Video, Q&A, and Slides

Missed the Arun 'Triple A' event or want to revisit it? We've got you covered! Check out the video, Q&A, and slides now.

Community Summit North America

Community Summit North America

Innovate, Collaborate, Grow. The top training and networking event across the globe for Microsoft Business Applications

Top Solution Authors
Top Kudoed Authors