Basic DB setup

Question

Hi,
I've just started to look into Oracle 11g OLAP cubes and wondered what the best way to structure my DB was. I have the following situation:
1. Hierarchy of Country > City
2. Hierarchy of Business > Branch
3. Sales in each branch
Is it best to set up the db fact and dimension tables as follows:
sales-fact
salesid PK
countryid FK to country-dim
cityid FK to city-dim
businessid FK to business-dim
branchid FK to branch-dim
salesMetric1
salesMetric2
salesMetric3
where country-dim, city-dim, business-dim and branch-dim all follow the following convention:
xxx-dim
xxxid
xxx
OR is it better to handle the hierarchy of country > city and business > branch as follows:
sales-fact
salesid PK
locationid FK to location-dim
entityid FK to entity-dim
salesMetric1
salesMetric2
salesMetric3
where location-dim:
location-dim
location_d
country
city
and entity-dim:
entity-dim
entityid
business
branch
I'm not sure which is the better method of contruction for use in AWM for generating cubes, where location and business would be dimensions of the cube.
Thanks for all help

Answer 1

The second approach, with separate dimension tables, is much better for many reasons.
(1) The dimensions will load faster since you will be scanning a smaller table;
(2) The fact table will be smaller since it has less data, so the cube will also load faster;
(3) You are less likely to have bad hierarchy data since there is less duplication;
(4) You have more flexibility to modify dimension hierarchies if and when your needs change.
This is basically a standard star schema, which is about as good as it gets from AWM's perspective.