Pardon me while I work thru some very basic issues of data model design. What I’m dealing with is a basic contradiction. On the one had, Rob Collie suggests that I should keep my columns in my primary data table to a minimum. But, cube design basics suggest that I should lean more toward the star model than the snowflake. I can’t have it both ways.
Here is a snowflake which requires only one reference column in my primary data table:
And here is the star which requires three reference columns in my primary data table:
So which is better? I’m gonna go with the star here, in spite of the additional reference columns. Why? Because a star, pretty much by definition, means using more reference columns so when they say “use a star” they must necessarily mean “go ahead and use more reference columns, it’s worth it.”