MDAC 2.5 SDK - OLE DB Programmer's Reference
Chapter 23: Rowsets in OLE DB for OLAP

Using Measures on the COLUMNS Axis

In the special case of a dataset consisting of only the Measures dimension along the x-axis, the resulting flattened view is typical of the way in which atomic multidimensional data is commonly stored in a relational database. This table of granular data is called a fact table. The combination of a fact table and a set of related dimension tables make up a star schema configuration.

For example, consider the following MDX statement. (Unqualified names are assumed to be unique.)

SELECT 
   {Sales, Cost} ON COLUMNS,
   CROSSJOIN({USA, Asia}, {Qtr1, Qtr2, Qtr3, Qtr4}) ON ROWS
FROM SalesCube
WHERE ([1991], Computers)

This produces the following dataset:

In turn, this dataset yields the following flattened rowset:

This representation is intuitive and corresponds to the way in which such data is usually displayed in a tabular view.

This example also illustrates step 2 of the algorithm in the section "Flattening Algorithm": If the DIMENSION PROPERTIES clause is not specified, it is equivalent to specifying DIMENSION PROPERTIES MEMBER_CAPTION.