Hello,
We have multiple date dimensions for a cube. For instance, values like
Start Date, Ship Date, Receive Date, etc.

The way this is currently set up is strange. Multiple date dimensions
were created for each date value so we have dimensions such as Start
Date, Ship Date, Receive Date, etc. BUT they are based off multiple
tables, identical copies of the original date table.

Wouldn't it be better just to have the multiple dimensions, but base
them off the same table? Would that cause problems in processing the
cube?

thanks for any advice,
Sylvia

In my own experience, I've either created multiple views of the SQL
table; or, as you are suggesting, created multiple dimensions from the
same SQL table. A couple of issues to keep in mind with the second
option:

- Aliases for the table are needed in the cube schema, with an instance
for each dimension that uses the table.

- In my case, drill-through to details didn't work as expected for the
cube (don't if this is still an issue).


Pl. note that AS 2005 (Yukon) allows use of a dimension in multiple
roles, which is a simpler solution:

http://www.microsoft.com/technet/pro...uate/dwsqlsy.m
spx
...
Analysis Services 2005 adds important new dimensional structures. Some
of these structures have odd names, but these names are common in the BI
literature.

• Role Playing: The dimension plays several roles depending on context.
For example, the [Time] dimension may be reused for [Order Date] and
[Ship Date]. In 2005, the role playing dimension is stored once and used
multiple times. Disk space and processing times are minimized.

- Deepak

Deepak Puri
Microsoft MVP - SQL Server

*** Sent via Developersdex http://www.developersdex.com ***