Hi all,
Some time ago we had a team of developers who built a data warehouse.
It featured very distinct data marts: Finance, car parts and cars.
However as the business has evolved there is a need to join these. How
is this done? Do we need to design new tables that allow a join or do
we need to redesign the whole data base structure?
Thanks
Sam

Bring all the fact tables & dimension tables together in a relational
database (e.g. SQL Server).
Identify which dimensions can be shared.
Create the shared dimensions (e.g. in Analysis Services).
Create the cubes from the fact tables.
Ensure you have a good front-end app. that can bring data from more
than one cube into the same report using the shared dimensions.
John Keeley

www.johnkeeley.com

Bring all the fact tables & dimension tables together in a relational
database (e.g. SQL Server).
Identify which dimensions can be shared.
Create the shared dimensions (e.g. in Analysis Services).
Create the cubes from the fact tables.
Ensure you have a good front-end app. that can bring data from more
than one cube into the same report using the shared dimensions.
John Keeley

www.johnkeeley.com

Hi John,
I saw this posting too and wonder if you could expand?
What do you mean by "shared dimensions" do you mean the formation of new
tables which hold both the new and old attributes?
thanks


duvinrouge (AT) servihoo (DOT) com wrote in message
<1106740252.696902.102080 (AT) z14g2000cwz (DOT) googlegroups.com>...

Presumably Sam has a fact table for cars & a fact table for car parts.
Perhaps both fact tables have a column for "make & model".
A dimension table is created with all the make & model members.
This dimension table is then used in both star schema.
If using Analysis Services then you will create a shared dimension for
make & model.
And a cube for cars & a cube for car parts.
If you have a good front-end app. then you can build a report that
takes numbers from both cubes & has the slicer dimension make & model
to slice your report.

Regards,

John Keeley

www.johnkeeley.com