Databases Reference
In-Depth Information
a state's country won't change over time. On the other hand, click the
customer address relationship and you will see it is set to Flexible
because a customer's address can change over time.
Figure 13-37
The aggregation design algorithm bases the aggregation being designed on
the type of relationships between the attributes it's using, allowing you to clas-
sify aggregations as rigid or flexible. Rigid aggregations are aggregations that
include attribute(s) that have a rigid relationship with the granularity attribute.
Attributes from the remaining dimensions either need to be the All level or the
lowest level. The aggregation created in the previous section that included
Customer.Country, and all other dimensions where the other dimensions in-
cluded the top level is an example of rigid aggregation. Flexible aggregations
are aggregations that are built on one or more attributes with flexible relation-
ship with the granularity attribute. An example of a flexible aggregation is an
aggregation that uses Customer.Address attribute.
Rigid aggregations are updated when partitions are incrementally processed.
If attributes that are part of rigid aggregations are incrementally processed
then existing aggregations are not dropped. Analysis Services will keep the
old aggregation data as such and create a temporary aggregation store for
the newly coming data. Finally Analysis Services merges the temporary ag-
gregation store with the old aggregation data within a transaction. Old ag-
gregation will still be available for query access when aggregations are being
Search WWH ::
Custom Search