Are there DAX implications for using primary keys in dimension tables?

[Anonymous]

I have a PBI data model, which is a star schema, but am wondering if there are any DAX implications for the design (I know certain data model designs can cause undesired results when writing DAX measures, e.g. auto-exist in conjunction with a single table model).

My fact table in PBI looks like this (there are a few more columns); it's a database view:

development_PK   development_spend
DEV0001          £1000
DEV0002          £3000
DEV0003          £2000

To bring in a new dimension (something I want to filter or slice by), I create a view in the database that has the development_PK and the relevant value from the location table:

development_PK   location
DEV0001          UK
DEV0002          France

I'll then join on development_PK in the PBI data model.

I could, of course, bring the primary key from the location table into the view I'm using for the fact table, where it would be a foreign key, and then simply connect to the location table.

[amitchandak]

@Anonymous , if this direct query then it is not a good option. But if it import mode, a view with distinct is fine. You can do that in power bi too.

refer: https://www.youtube.com/watch?v=kU2M1LmNvNo

DAX

https://www.youtube.com/watch?v=vYxWyt0qNUQ

[Anonymous]

It's import. What do you mean what when you say: 'a view with distinct' please? Thanks.

PS I know how to create the tables in Power Bi. I'm asking if the dimension not including it's own Primary key is an issue.

[v-kelly-msft]

Hi  @Anonymous ,

It is a good practice if the dimension includes it's own Primary key because it helps ensure that all your facts reference existing dimension members, and may improve performance but it is not mandatory.

Best Regards,
Kelly

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