Exluding lowest level dimensionmembers from a total report.

Question

Hi,
I know this thread might belong in the Discover part of the forum, but based on previous experience it seems that Discoverer for OLAP help is difficult to come by in that part. Also I think this quickly might end up with being a bit more OLAP technical.
To the issue:
Say that you have a sales cube with 3 dimensions, time, product and customer.
The customer dimension has three levels with one standard hierarchy in the following order: Total, customerclass and customertype.
For the lowest level there is a customertype member thats named "internal". This means that all employees buying from their own company with a special discount is placed under this member.
Now to the report:
Using discoverer for Olap, management want two total reports. One where the Internal member is included, and one where it is excluded. This meaning that the initial report should just show the total levels for all dimensions, just with different totals.
Now I thought of just not include the internal member when you choose different dimensions in the query builder, but as far as I know this would only work for the lowest level. Total and customerclass would still show sums that include the internal level since the cube has aggregated these levels in advance.
Does anyone a way to work around this?
some cube facts:
database version 10.2.0.3. with the OLAP A patch
cube compressed and partitioned along month, fully aggregated for all levels(except year).
regards Ragnar

Answer 1

Rangar,
Simplicity is in the eye of the beholder, but you should be able to make it work. The details will depend on your source mapping, but I will assume that you have a single dimension star table with three columns, one per level.
CREATE TABLE CUST
Total varchar2,
CustomerClass varchar2,
CustomerType varchar2
I'll assume further that the member 'INTERNAL' appears as CustomerType. Define a VIEW on top of the table like this.
CREATE VIEW CUST_VIEW AS
SELECT *
FROM CUST
WHERE CustomerType != 'INTERNAL'
Now create two new levels
TOTAL_EXTERNAL
CUST_CLASS_EXTERNAL
Create a new level hierarchy, EXTERNAL_HIER say, that includes the two new levels along with the existing CustomerType level.
TOTAL_EXTERNAL
CUST_CLASS_EXTERNAL
CUSTOMER_TYPE
Note that even though the CUSTOMER_TYPE level appears in two hierarchies, the members in that level can vary be hierarchy. This is important since it means you do not have to reload the leaf data for the cube.
Now map the new hierarchy to the three key columns in CUST_VIEW. To do this correctly you need to switch to the "Snowflake Schema" view on the AWM mappings tab. This allows you to explicitly set the parent column for each hierarchy. So, in your new EXTERNAL_HIER hierarchy the parent of CUSTOMER_TYPE is CUST_VIEW.CustomerClass. In your old hierarchy, INTERNAL_HIER say, the parent of CUSTOMER_TYPE is CUST.CustomerClass. If you don't switch to the "Snowflake Schema" view then you will end up with null values for the parent dimension. (I just wasted 45 minutes discovering this fact.)
You now have two hierarchies and five levels. Your aggregation strategy for the cube should reflect this in some way. If you want full pre-aggregation, then you will need to add the two new levels to the cube. If not, then you are probably set.
David