Data Modeling Question

[a4apple]

Hello Everyone,

I am working with Orders and OrderItems model. There are 6 different individual dimension keys that make up a primary key including a date field. To effectively use the filters I am using the linking table concept (Created a Primary key in the linking table via ETL and using that table to filter the facts). 

Basically my model is like,

Orders <==> OrdersDimension <==> Rest of Dimensions

OrderItems <==> OrdersDimension <==> Rest of Dimensions.

/t5/image/serverpage/image-id/3095i6752181269DACC47/image-size/large?v=1.0&px=600

Works fine, but with historical data, I have more than 2Billion unique PK records via ETL which is a limitation in Tabular.

How can I get around this limitation?

I have tried another solution like linking the dimensions to the fact tables directly. but I have to write complex DAX to propagate filter conditions to OrderItems measures.

https://stackoverflow.com/questions/10181316/2-billion-distinct-values-per-column-limitation

Basically to solve a problem like this

https://community.powerbi.com/t5/Desktop/Best-Way-for-work-with-Multiple-Fact-Tables/td-p/21441

I saw an answer provided in this solution but I have no clue how to merge?

Any suggestions are appreciated. Thanks in advance.

[GilbertQ]

I would suggest looking to aggregate your data, as often when I have been working people do not go down to the very granular data.

This will ensure that your model is smaller, as well as possibly looking to merge some of the data into smaller tables?

[a4apple]

Thank you @GilbertQ . At this moment that is one of the options I am considering. I am also trying to understand if there is any other work around for this situation meanwhile.