I'm trying to bootstrap myself into an understanding of dimensional database design, so forgive.....

One of our requirements is that an analytic user group be able to add a new attribute to an analysis database within an hour.

They have a SQL-savy analyst who can modify a table and add the requested new data.

I'm thinking of creating a primary customer dimension table that contains the previously agreed-to data elements. Our production group would load that table on a periodic basis (quarterly) when the source data is refreshed. I'm also thinking of creating an "auxiliary" customer fact table that this user analyst could modify and update as needed. This way we'd receive both the benefits of the production processes our development team has put in place, while providing the user group the flexibility and responsiveness they require.

Does this sound like a good (or acceptable) approach?

Nolan

I think it is definitely a good idea to keep user-implemented and untested schema modifications as separate from pure application schema as possible....