Partition Structure
The structure of a partition must match the structure of its cube. The set of and that defines a cube's structure must also be present in the partition. For this reason, when a partition is created, it automatically inherits this set of measures and dimensions from its cube; it is not possible to edit this set within Analysis Manager.
However, it is possible for a partition and its cube to have different and/or . Different partitions in a cube can have different fact tables and/or data sources. In these cases, a high degree of similarity among the fact tables and data sources of the partitions and the cube is required to ensure that the matching structures of the partitions and their cube are not undermined.
A cube and all of its partitions must have fact tables with the same structure. All columns used in the cube's measure definitions must be present and must be named the same in all of the fact tables. Extra columns can be present in some fact tables but absent in others. The fact tables can have different names.
All of the data sources of a cube and its partitions must reference databases that contain a identical to the schema in the definition of the cube. (You can display this schema in the Cube Editor Schema tab.) All dimension tables in the schema of the cube must be present and must be named the same in all of the databases. A fact table meeting the requirements described earlier in this topic must be present in each of the databases. All columns that join tables in the cube's schema must be present and must be named the same in all of the databases. The dimension tables must have the same structure in all of the databases. All columns used in the level definitions and member property definitions of the cubes must be present and must be named the same in all of the databases. Extra columns can be present in some databases but absent in others. Extra tables outside the required schema can be present in some databases but absent in others.
See Also
Cube Structure