Measures and Multidimensional Analysis - Microsoft SQL Server 2008 Analysis Services - page 83

</Dimension>
</Dimensions>

Measure Group Dimension Attributes and Cube Dimension
Hierarchies

Take a look at Figure 7.5. Suppose that you change the granularity attribute for the
Customer dimension in our sample from the Customer attribute to the City attribute.
You'll lose sales data not only for the Customer attribute, but also for the Gender attribute.

Attribute Tree Hierarchy

Geography

Native
Hierarchy

Country

Country Country

State Province Province

City City

City Gender

Customer Customer

Customer

FIGURE 7.5 The key attribute Customer is used as the granularity attribute and in the user-
defined hierarchy Geography.

When the granularity of a measure group changes, it's not easy to figure out which attrib-
utes are related and which are unrelated to the measure group. To make a decision,
Analysis Services uses dimension hierarchies defined for a cube. By default, when the gran-
ularity attribute is the key attribute, any cube dimension hierarchy can be used to analyze
measure group data. When an attribute other than the key attribute defines the granular-
ity of a measure group for a dimension, only the hierarchies that pass through this
attribute are available for analysis. For a precise definition of which hierarchies and attrib-
utes are available for analyzing measures of a measure group, we introduce a new term:
native hierarchy. A native hierarchy is a foundation for every dimension hierarchy. It
includes all the attributes that are included in the dimension hierarchy and any attributes
that are part of the uninterrupted path, through attribute relationships, from the top
attribute of a hierarchy to the key attribute. If there is no uninterrupted path, the hierar-
chy is an unnatural hierarchy. When it encounters an unnatural hierarchy, Analysis
Services divides it into two or more native hierarchies. Analysis Services builds one native