I'm new to modeling star schemas, fresh from reading the Data Warehouse Toolkit.
I have a business process that has clients and non-clients calling into conference calls with some of our employees.
My fact table, call it "Audience", will contain a measure of how long an attending person was connected to the call, and the cost of this person's connection to the call. The grain is "individual connection to the conference call".
Should I use my conformed Client dimension and create a non-client dimension (for the callers that are not yet clients) this way (omitting dimensions that are not part of this questions):
Or would it be OK/better to have a non-conformed Attending dimension related to the conformed Client dimension in this manner:
Or is there a better/standard mechanism to model business processes like this one?
Edit:
What about using model 2 above, but creating a view on top of the client dimension table and the attending dimension to make it look like it is only one dimension?
Is that an acceptable alternative to Damir's answer below?