IBM Cognos User Manual page 189

To speed up cube creation, reduce cube size, or improve run-time access for your
OLAP report users, you may need to
v adjust your data
v streamline your data sources
v redesign your model
v select different cube processing options
Tip: When you make changes to your model, be aware that the changes you make
may impact report authors and report consumers. For more information, see
"Member Unique Names" on page 22.
Consider making the following improvements:
v Delete records from the source data if they are no longer needed or are out of
date. Consider excluding data from the cube, or consolidating the data by
summarizing the details.
v Use multiple data sources where feasible, to reduce the size of each source file
and shorten processing time in Cognos Transformer. For more information, see
"Use Multiple Data Sources" on page 177.
v Subdivide large PowerCubes into multifile cubes to improve run-time query
performance.
v Ensure that structural data sources (those that contain the hierarchical data for
your dimensions) appear first in the Data Sources list, followed by the
transactional data sources (those that supply the measures for your model).
Include only the minimum number of columns needed to reference your
dimensions.
v If you specify a unique level, ensure that your data does not have uniqueness
violations. Allocate extra time for data source processing to verify that all
categories are unique within a level, or eliminate this step if it is not necessary
by clearing the Verify category uniqueness option on the property sheet for the
data source.
v Consolidate transactional data to a higher summary level using the Duplicates
Rollup feature, and implement optional drill-through targets if the details are
not needed by the majority of your users.
You can also select one or more of the following options from the Processing tab of
the PowerCube property sheet.
Optimization
Use this option to specify how cube creation is optimized. Auto-partition is the
preferred optimization method.
The optimization methods include:
v Auto-partition
Auto-partition is the default optimization method. With Auto-partition, you can
let Cognos Transformer determine where to place the partitions or you can
define the partitions manually.
Some situations may prevent the use of Auto-partition optimization. In these
situations, Cognos Transformer uses Categories optimization. For example, if a
model has two cubes, one with optimization set to Auto-partition and the other
with optimization set to Categories, Cognos Transformer will build both cubes
using the Categories method. This is because Cognos Transformer cannot use
Chapter 8. PowerCubes in Production Environments
175