I have encountered an odd and interesting model in my travels. It uses what I call 'tradeoff' dimensions. Imagine a case in which you have a full P&L and BS in your accounts dimensions. Additionally, you have statistical accounts that number in the hundreds. It is possible to configure a model which splits all of the metrics in half - instead of integrating them into a single accounts dimension, you can split them into tradeoff dimensions. You select "none" in one in order to get data from the other.
I am not aware of the particular circumstances under which this design was put together and accepted, but there it is.
It certainly sounds like a great case of dimensional irrelevancy.
Okay, I'll bite -- were both of these dense (I am guessing a BSO db)?
Having said that, I have seen an Essbase app (ASO if that matters) that in fact has two "Accounts" dimensions -- one to show data pre transform, the other to show post transform. But at the end of the day, it was the same number, just in different hierarchies (and oh yes, in separate Cost Center/Profit Center dimensions). It sort of surprised me at first, but it makes sense in the context of the app.
Regards,
Cameron Lackpour
Posted by: Cameron Lackpour | February 25, 2010 at 07:58 AM