Many of the problems associated with the basic star schema are resolved with the BW extended star schema. With the extended star schema, attributes are removed from the dimensions and placed outside the InfoCube in master data tables.
The BW extended star schema differs from the basic star schema. It is divided by a solution dependent part (InfoCube) and a solution independent part (attribute, text and hierarchy tables) which is shared among other InfoCubes.In BW, attributes located in the dimensions are called characteristics. In BW, attributes located in a master data table of a characteristic are called attributes of the characteristic. When designing a solution, it is a great challenge to decide whether an attribute should reside in a dimension table and thus in the InfoCube or in a master table or even both. Data is loaded separately into the master data (attributes), text and hierarchy tables. The SID table provides the link between the master data and the dimension tables.
The fact table and the relevant dimension tables of an InfoCube are connected with one another relationally using the dimension keys. The dimension key is provided by the system per characteristic combination in a dimension table.
With the execution of a query the OLAP processor checks the dimension tables of the InfoCube to be evaluated for the characteristic combinations required in the selection.
The dimension keys determined in this way point the way to the information in the fact table. Dimension tables consist of a maximum of 248 characteristics. The Time dimension holds the time characteristics needed for analysis. The Unit dimension contains the unit of measure and currency characteristics needed to describe the key figures properly. The Data Packet dimension is used to identify discrete packets of information loaded into the InfoCube. In this way, packets can be deleted, reloaded or maintained individually.