IBM Cognos Dynamic Cubes

90 IBM Cognos Dynamic Cubes

4.9.9 Automatic aggregate cube creation

You can use the aggregate cube auto-matching functionality to create an aggregate cube

automatically. You can do this by dragging the aggregate table onto the aggregates field of the

aggregates tab of the cube that you are working on. Where matching measures and

dimensions are found in the cube, Cognos Cube Designer maps each of these items to the

aggregate table. Where possible, it also attempts to identify the highest level of aggregation

required and roll up dimensions. Inspect the aggregate cube and refine it to ensure that it is

modeled exactly as you need it to be.

The ability to automatically map depends on how the aggregate tables are set up. A naming

convention for aggregate cube columns that conforms them to their analogue source objects

is helpful. Complete the following steps:

1. Select the gosldw_sales cube.

2. Expand the Project Explorer tree node for the cube or right-click Open Editor and click

the Aggregates tab.

You see an aggregate cube called gosldw_sales2.

3. Expand the metadata tree in the Data Source Explorer and select

AGGR_TIME_PROD_OM_FACT.

4. Drag AGGR_TIME_PROD_OM_FACT onto the Aggregate Cube Editor.

5. A new aggregate cube is created. Examine the dimensions, measures, levels, and level

key mappings that are created in the aggregate cube.

Cognos Dynamic Query Analyzer has an Aggregate Advisor, which can identify, from the

model or from log files, possible candidates for the creation of aggregate tables. For more

information about the Aggregate Advisor, see Chapter 8, “Optimization and performance

tuning” on page 201 and the Cognos Dynamic Query Analyzer documentation.

4.10 Validation

Cognos Cube Designer can monitor you, as you model and analyze the model. The validation

checks for model integrity and attempts to assist the modeler to follow modeling best

practices. The results of this validation are displayed in the Issues tab in the property view. If

issues are identified, an icon opens on the object that has the issue and also all parent

objects. This feature helps you ensure that your chances of seeing that an issue has been

generated are good.

There are issues with a severity of error and issues with a severity of warning. The former

prevents the modeler from being successful with the model. The latter is not necessarily a

problem but, as the modeler, you must attend to it and verify whether you must make a

modeling change.

Some examples of the model integrity validation checks include identifying if levels do not

have attributes in them, identifying if levels are not used in a hierarchy, identifying if an

attribute does not have an expression or object defining it. Other tests include checking to see

if the relationship between a dimension and the measure dimension are defined.

One test of modeling best practices is to ensure that the relationship cardinality flows down to

the fact table. If an object that has a relationship to another object in the dimension goes

against this flow, Cognos Cube Designer flags it as an issue.