Join us at FabCon Atlanta from March 16 - 20, 2026, for the ultimate Fabric, Power BI, AI and SQL community-led event. Save $200 with code FABCOMM.
Register now!To celebrate FabCon Vienna, we are offering 50% off select exams. Ends October 3rd. Request your discount now.
All - I have a test data model design here where I have multiple normalized dimension tables but the measures don't seem to give me the right counts. When you have many dimension tables like this that have many to many's, is this the way to properly setup the data model, or is it better to try to design a different, denormalized design?
Don't attempt to join fact tables. They should only be controlled by common dimensions. Don't have 1:1 relationships. Don't have bidirectional relationships unless it is absolutely unavoidable.
@lbendlin I got rid of the fact to fact joins. The 1:1 i will merge together. But is there any better option than still having this long string of bidirectionals to span all the factless fact and dimension tables?
As long as it remotely resembles a star schema you should be good. The data model needs to follow the business question, don't expect a single data model version to anwer all of yours.