Understanding Cubes

A cube is an OLAP metadata entity. This means that it is a logical object, identified by name and owner, within the OLAP 2 Catalog.

OLAP cubes represent dimensioned data. Cubes must exist for any data that you wish to make accessible to the Oracle OLAP API.

Cubes and Measures

A cube is a dimensional framework to which you can assign measures. A measure represents data stored in fact tables. The fact tables may be relational tables or views. The views may reference data stored in analytic workspaces.

A measure's data can be accessed by specifying values for its dimensions. For example, a measure representing sales data might be dimensioned by time, product, and location. This means that the sales data can be accessed for a given time period, a given product, and a given location.

Note: A cube may contain multiple measures. All the measures of a cube share the same set of dimensions.

Fact Table Requirements

The fact tables or views that underlie a cube have data columns and key columns. The data columns store the source data for the cube's measures. The key columns reference the dimension tables that underlie the cube's dimensions.

If your measures and dimensions are stored in an analytic workspace within the database, you must call procedures in the AWCONVERT package to create the fact tables as relational views that reference the workspace. Then you can call procedures in the CWM2_OLAP_CUBE package to create cubes in the OLAP 2 Catalog.

Fact Table Key Columns

Each fact table key column references a level column in a dimension table. The level is mapped to one of the dimension's hierarchies. This means that fact data is stored on a per-hierarchy basis.

Note: The measures for a single cube may be stored in more than one fact table. The key columns of each fact table may store different hierarchies for the same dimensions.