Hyperion OLAP model vs. Oracle OLAP

Question

Hi, I have a question re: how to properly model stuff in Hyperion OLAP. I know the basics of Essbase, but I've never found out how to do this...
In Oracle OLAP, I can define a set of dimensions, and then create cubes that use any subset of those dimensions.
In Essbase, if I define a set of dimensions, there is only one cube - and it is forced to share ALL of the dimensions.
So I'm always at a loss how to model something that would be pretty simple in Oracle OLAP, like:
Sales cube - dimensioned by time, store, product, customer
Local tax rate cube - dimensioned by time, store (but not product or customer)
Sales goal cube - dimensioned by time, store, product (but not customer)
etc.
What's the best way to handle this in Essbase?
p.s. after these cubes are done, I'll also want to do some formulas like local taxes = sales cube * local tax rate, etc.
Thx,
Scott

Answer 1

Essbase basically has two different design approaches. One you build a cube via Admin services and or load rules, the other is via Integration Services. When using Integration Services, it fits the more traditional build methodology you are familiar with. Via Integration Services you would define a star schema with a fact table and dimensions. You would then build a meta-outline layer where you would define how you want the cube to be built. You could "publish" a cube with all dimensions, or you can have multiple versions of your meta-outline, some that use certain dimensions and others that do not. Within that model you could go one step further and apply filters so you could build two cubes with the same dimensions but filter groups of members out of the dimension in one cube that you allows to flow through in the other cube, or maybe only allow it to build down to a certain level. The options get pretty broad. I think EIS will will get you to what you are familiar with and make a lot more sense. Go to the download site and you can download the docs on EIS to start to get a better idea.