Advanced Modeling - Microsoft SQL Server 2008 Analysis Services - page 97

<NullProcessing>UnknownMember</NullProcessing>
<DataType>Integer</DataType>
<Source xsi:type=”ColumnBinding”>
<TableID>salary_by_day</TableID
<ColumnID>employee_id</ColumnID>
</Source>
</KeyColumn>
</KeyColumns>
<Type>Granularity</Type>
</Attribute>
</Attributes>
</Dimension>
<Dimension xsi:type=”ReferenceMeasureGroupDimension”>
<CubeDimensionID>Store</CubeDimensionID>
<Attributes>
<Attribute>
<AttributeID>Store</AttributeID>
<Type>Granularity</Type>
</Attribute>
</Attributes>
<IntermediateCubeDimensionID>Employees</IntermediateCubeDimensionID>
<IntermediateGranularityAttributeID>StoreId</IntermediateGranularityAttributeID>
</Dimension>
</Dimensions>
</MeasureGroup>
</MeasureGroups>
</Cube>
It's possible to get to a point where you have a chain of referenced dimensions, when a
dimension that is bound to the fact through another dimension is itself a referenced
dimension, and so on and on. However, you need to be careful about creating loops of
referenced dimensions.
NOTE
The creation of chains of indirect referenced dimensions is not supported by Analysis
Services.
Indirect referenced dimensions don't define how data is loaded into the fact, making it
possible to change a dimension, or change whether the dimension is included in the
measure group, without reloading the data. Members of the dimension are resolved at the
time of the query, rather than upon loading the data. Even if you completely change the
data of the dimension or change the structure of the dimension, the system won't have to
reload the data. The system also doesn't index the measure group data by the indirect
referenced dimension. Therefore, the performance of queries working with indirect refer-