AWM: Enable Analytic Workspace for OLAP API

Question

Hi,
I created an analytic workspace with Analytic Workspace Manager which is based on a relational star-schema.
Now I want to access this analytical workspace by normal SQL-Statements. For this I executed in AWM the 'Enable Analytic Workspace for OLAP API and BI Beans ...' routine which generated views that contain the fact and dimension data.
My problem is that these views are very huge, because they
apparently store all possible combintations of dimension- and fact records. Most of the records contain only NULL values in the fact columns. I also used Composite Dimensions for the fact table in order to reduce sparsity.
Is it normal to have such huge views with mostly NULL values? If yes, how can I handle this in order to get an acceptable performance.
Here an example of a small fact table I use:
A fact table with 5400 records with its two dimension tables (1129 and 233 rows, no hierarchies) produces 263057 rows in the view.
Thanks for any help.

Answer 1

Hi,
First of all, thanks for the quick response flavio.
Meanwhile I have found a solution to my problem:
I don't use the 'Enable Analytic Workspace for OLAP API and BI Beans ...' routine of AWM anymore. Instead, I write my own statement for creating a relational view for the analytical workspace. The statement looks like following:
create or replace force view costrate_cube as
SELECT producttype_number, producttype_description, machine_sk, notok_costrate, production_costrate
FROM TABLE(CAST(OLAP_TABLE(
'costrate_aw DURATION SESSION',
'costrate_table',
'FETCH producttype_dim_producttype_number,
producttype_dim_producttype_description,
produnit_dim_machine_sk, produnit_dim_module_sk,
notok_costrate_variable, production_costrate_variable
USING <costrate_cube_composite>','') as costrate_table));
The important part is the USING clause of the FETCH command. By using the composite of all dimensions of the cube, rows containing only null values in the fact columns are being eliminated. Now I get only the rows which are in my original relational fact table (which is what I wanted).
Werner