Dimensionally modeled data, whether stored in cubes or stored as Dimensionally Modeled Relational (DMR) data, organizes data into dimensions. These dimensions contain hierarchies. The hierarchies contain levels. And the levels contain members.
An example of a dimension is Locations. A Locations dimension may contain two hierarchies: Locations by Organization Structure and Locations by Geography. Either of these hierarchies may contain levels like Country and City.
Members are the instances in a level. For example, New York and London are members in the City level. A member may have multiple properties, such as Population, Latitude, and Longitude. Internally, a member is identified by a Member Unique Name (MUN). The method by which a MUN is derived depends on the cube vendor.
Relational data models are made up of data subjects, such as Employees, which are made up of data items, such as Name or Extension. These data items have values, such as Peter Smith.
In ACI, the methods of drilling through available are
Dimensional (member) to Dimensional (member)
Dimensional (member) to Relational (data item value)
Relational (data item value) to Relational (data item value)
If the target parameter is a member, the source must be a member. The source and target should usually be from a conformed dimension. However, if the data will support it, you may also choose to define a mapping using different properties of the source metadata item.
If the target parameter is a value, the source can be either a value or a member. If the source is a dimensional member, you must ensure that the level or dimension is mapped to the target data item correctly in the drill-through definition. The business key from which the member is sourced should usually match the relational target value, which is most often the business key. However, if the data will support it, you may also choose to define a mapping from the caption of the source metadata item.
