IBM Cognos, User Manual

Page: 190 / 411

two different optimization methods at the same time. The Check Model feature
issues a warning indicating when Categories optimization is used instead of
Auto-partition.
For information about cube partitions, see “Choosing a Partitioning Strategy” on
page 180 and “Partition Manually (if Required)” on page 182.
v
Categories
This is an older optimization method that is used by Cognos Transformer when
cubes are incrementally updated. The Categories method can handle certain
model conditions, such as a before-rollup calculated measure or selecting no
consolidation when processing records for a cube.
If you want to partition the cube and use the Categories optimization method,
you must set the partitions manually in the cube. More than one level of
partitioning will increase the cube build time substantially.
v
Data passes
The data passes optimization method is a variation of the categories method.
When Cognos Transformer uses the data passes method, the number of passes
through the temporary working files created during cube creation is reduced. All
categories are placed in the cube, however those with no measure data attached
are hidden when the cube is read.
With the data passes optimization method, the resulting cube may be larger and
cube read processing may be slower. This is because duplicate data points are
not consolidated, and all categories are included in the cube.
v
Direct create
The direct create optimization method is also a variation of the Categories
method. When Cognos Transformer uses the direct create method, all categories
in the model are added to the cube while the data sources are being processed.
Records that do not generate new categories are directly updated to the cube.
Use this method only when your model is expected to generate few new
categories, and when all categories are added to the cube.
You cannot use the direct create method for individual PowerCubes in a cube
group.
This Cube is Incrementally Updated
Use this option to add new data to an existing cube. If your records change only
slightly, this option reduces the time needed for cube processing.
Alternatively, you can create data sources that contain only the new data, and use
these sources to incrementally update your cube.
Remember to keep your model synchronized with your incrementally updated
cube and, even if your model does not change, recreate your cube periodically in
its entirety. If you do not recreate your cube periodically, and both the frequency
and number of incremental updates is high, processing performance may
deteriorate over time. For more information, see “Update Cubes Incrementally” on
page 179.
Cube Creation
Use this option to create the cube or cube group locally. The default is Enabled . If
data related to a particular cube is unchanged since the last update, shorten
processing time by selecting
Disabled .
176
IBM Cognos Transformer Version 10.1.1: User Guide