Keep in mind that a view still carries the inherent cost of the SQL lying under it - every time you process the cube. There are other optimization costs, as well.

What you're doing is really no different than graphically designing the cube based upon a relational schema, if I understand your description of the steps you've taken. "Anything that works" might be fine as an experiment, to gain a feel for how the application can be used across a host of database designs and schemas, possibly even as an exercise in a "basics" book, etc., but the same concepts apply within the star schema vs. OLTP optimization arguments.

Hope this isn't destined for Production in any way ...

Bill

William E. Pearson III
Island Technologies Inc.
931 Monroe Drive
Suite 102
Atlanta, GA 30308

404.872.5972 Office
888.570.8628 24 / 7 Messaging
404.873.3128 FAX

wep3@msas-architect.com
www.msas-architect.com