I have a fact tables with a composite primary key of 10 key columns.

6 out of 10 of the key columns are short, and with low cardinality, and also "mostly" unrelated to each other.

I am planning to move them out to a single junk dimension, so I can substitute the original 6 column with just one dimension ID in the fact table. My fact will have much tighter composite primary key, consisting only of 5 key columns.

My three short questions related to this design are:

  1. I have used junk dimensions before for flags, but only for non-key attributes. Is it something unorthodox to have a junk dimension ID participating fact tables primary key?

  2. I am not familiar with Power BI and Cubes. Does this junk-key-dimension setup would hurt the slicing-dicing in such data visualization tools?

Thank you!

  • I'm thinking Question 3 would be better off as a separate post Commented Nov 24, 2021 at 21:37
  • @MichaelKutz removed #3 just to focus on the junk dimension key aspect. Thanks.
    – Avi
    Commented Nov 25, 2021 at 7:44


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.