Cube Design - Professional SQL Server Analysis Services 2005 with MDX - page 237

The cube dimensions and measure groups are represented in a matrix format
as rows and columns, respectively, where the relationship between them cor-
responds to the intersection cell. The intersection cell shows the dimension
type along with the attribute that is used in the relationship to join.
Dimension Types
Six different relationships can exist between a dimension and a measure
group: No Relationship, Regular, Fact, Referenced, Many-to-Many, and Data
Mining. In Figure 6-13 you see two of the six relationship types; No Relation-
ship and Regular. Cells corresponding to a specific dimension and measure
group can have an attribute specified that indicates that the dimension type is
Regular. Further, such attributes can be used in the join condition between
the dimension and the measure group. Often this attribute is the key attribute
of the dimension and is called the granularity attribute. The granularity attrib-
ute can be an attribute that is above the key attribute of the dimension. When
you browse a dimension along with measures of a measure group where the
dimension and measure group have a regular relationship then Analysis Ser-
vices aggregates the data appropriately. The relationship between Dim Cus-
tomer and Fact Internet Sales measure group is a regular relationship. The
granularity attribute is shown in the cell intersecting the dimension and meas-
ure group as shown in Figure 6-13 .
Cells that are shaded gray indicate there is no relationship between the di-
mension and measure group. Whenever there is no relationship between a
dimension and measure group, the measure group property IgnoreUnre-
latedDimension controls the results of queries involving any hierarchy of that
dimension and any measure from the measure group. The measure values
will either be null (IgnoreUnrelatedDimension=False) or the same value for
each member of the dimension (IgnoreUnrelatedDimension=True). For ex-
ample, there is no relationship between dimension [Dim Employee] and the
[Fact Internet Sales] measure group. If you browse the Gender hierarchy of
[Dim Employee] and measure [Internet Sales Amount] you see that the meas-
ure values for each member of Gender hierarchy are the same value as the
Grand Total as shown in Figure 6-14 . This is because the IgnoreUnrelatedDi-
mension value is set to True by the Cube Wizard as a default. You learn more
about properties of measure groups and measures later in this chapter.