If we peek into data processing in big companies, we will find operational databases, which store business data in relational models. Business transactions are tracked online on these OLTP systems.

Regularly (e.g. weekly, daily) core data from the corporate OLTP systems are extracted from the OLTP systems and merged into a data warehouse (DWH). These data are stored in a multidimensional database, which is the basis for OLAP. The structures, which contain the data are called info cubes (synonyms: hyper cube, data cube ...).

The content of an info cube is determined mainly by the analysis topic, the needed level of detail, the user groups, who have to access the cube, the intervalls in which the cubes need to be updated and other factors.

Subsets of the data warehouse are being distributed into data marts, which contain departmental information.