The Incremental Update
function enables you to add new members to a cube and update data.
Every time a cube update is done, the incremental aggregation data
is written to a new SPDE table (rack). The more cube updates that
have been completed, the more tables or racks there are that must
be read during a query, and the slower the access time can be.
Periodically, it can
be beneficial for performance reasons to combine all of the individual
tables into a single aggregation table. Aggregations are coalesced
(combined) in a separate job from adding data to a cube. Only aggregations
with more than one rack are coalesced. The COALESCE_AGGREGATIONS option
can be executed only for the most current generation of the cube.
In addition, only MOLAP aggregations (those aggregations created by
PROC OLAP) can be coalesced. The coalesce task is processed in-place.
A new generation of the cube is not created.
After the cube generation
is coalesced, it is no longer dependent on the aggregation data from
previous generations. If those generations have already been deleted,
then the old aggregation tables are also deleted.
You can coalesce all
the aggregations by using the COALESCE_AGGREGATIONS option in the
PROC OLAP statement. In addition, the
Coalesce wizard is available in SAS OLAP Cube Studio.