How to query a cube using SQL ?

Question

Hi I have a MOLAP cube say CubeA with dimensionA having Attribute A1 and dimensionB having attribute B1
the measure of cube is M1.
Any idea how i can query the cube using SQL dislaying both the dimensions with the measure interlinking them.
I just need a general overview of how to sql query a cube .. I have the dimensions and sttributes and name of the cube but i ineed to know how to query it... I was given an example
SELECT
t.time_long_description,
p.product_long_description,
c.customer_long_description,
f.sales,
f.sales_ytd
FROM
time_calendar_view t,
product_std_view p,
customer_std_view c,
sales_cube c
WHERE
t.dim_key = f.time
AND p.dim_key = f.product
AND c.dim_key = f.customer
AND t.level_name = 'CALENDAR QUARTER'
AND p.level_name = 'BRAND'
AND c.level_name = 'STATE';
but i cant understand it.. i cant figure out what are the dimensions and where are the attributes...
Any idea would great...

Answer 1

There's a nice tutorial here: http://st-curriculum.oracle.com/obe/db/11g/r1/olap/cube/querycubes.htm
Here are a few basics. Note that the link between the dimensions is the join between the dimension or hierarchy views and the cube.
* The primary key to every dimension and hierarchy view is DIM_KEY.
* The primary key to the fact table is dim1, dim2, dim3, so all joins are between dim_keys and dim1, dim2, etc.
* Dimension, hierarchy and cube views contain rows for both leaf (detail) and summary rows (that means cubes can return very, very large numbers of rows is you aren't careful)
* You typically want to query for the aggregate level data rather than use SUM ... GROUP BY. Let the cube do the work and let it manage calculations.
* Dimension views contain rows for all dimension member. hierarchy views contain rows for only members of the hierarchy.
* You typically select one of the long or short description columns to display data within an application.
Here's a commented version of your query:
SELECT
t.time_long_description, -- long descriptions
p.product_long_description,
c.customer_long_description,
f.sales, --measures
f.sales_ytd
FROM
time_calendar_view t, --hierarchy views
product_std_view p,
customer_std_view c,
sales_cube c --cube view
WHERE
t.dim_key = f.time ; -- joins between hierarchy and cube views
AND p.dim_key = f.product
AND c.dim_key = f.customer
AND t.level_name = 'CALENDAR QUARTER' -- Filters to ask for data at the quarter
AND p.level_name = 'BRAND' -- brand
AND c.level_name = 'STATE'; -- and state levels.
You want to have a filter on every dimension to avoid return all rows in the fact table for that dimension (remember, the cube contains summary rows). If you want to drop a dimension from a query (e.g., you don't want product in the query, intending for the query to be the aggregate of all products), filter to an all/total level.
e.g.,
and p.level_name = 'ALL PRODUCTS'
or
and p.long_description = 'All Products'
There are many variation of this query, but if you understand the basics you'll work your way there.
And, never do a select count(*) from a cube. Doing so will select the entire cube, detail and summary data.