Chapter 4. Modeling dynamic cubes 63
editors are available only in the Model Explorer window. The default state of the Model
Explorer is a list of all of the objects in your model. Double-clicking any of the objects in
this list will open the appropriate editor for the selected object.
Immediately below the Model Explorer are the Properties and Issues tabs:
? The Properties tab contains the various properties and expression definition of the item
that is currently selected and in focus. Many of these properties are editable.
? The Issues tab contains a list of any unresolved issues with the object currently selected.
These issues must be resolved before the object you are working on is valid. A cube
cannot be deployed to Cognos Connection if there are outstanding issues to be resolved.
Cubes
Dynamic cubes consist of a collection of dimensions, representing dimension tables in the
data source, and a single measure dimension, representing the fact table. There can be only
one measure dimension in a dynamic cube, but there can be any number of regular
dimensions in a cube. The term
regular dimension is used to distinguish between dimensions
that contain member data versus those that contain fact data or measures.
The modeling exercise is to create the cube by deciding the fact table you want to use in the
measure dimension of the cube, identify the dimensions you want, their grain to the fact table,
and establish relationships between them and the cube’s measure dimension.
Dimensions
Dimensions are metadata constructs. They guide the relational queries and the specification
and creation of the hierarchies and members of the dimensions.
A measure dimension is the container for the facts in the cube. These facts might exist in the
data warehouse or can be created as expressions. They can be values or created from
counts of events, states, and other attributes that your consumer might be interested in
tracking. The latter are commonly known as
fact-less fact tables.
Three types of dimensions exist in addition to the measure dimension: regular dimensions,
time dimensions, and parent-child dimensions.
Hierarchies
A hierarchy is a structure that organizes and relates the attributes of a dimension into layers
of increasing level of detail from the most abstract to the most concrete.
For example, a time dimension hierarchy can consist of a year level, a month level, and a day
level. The hierarchy’s highest level, year, has the most abstract information classifying time. If
you were examining something from the level of years, all the data would be aggregated at
that level. Months are more detailed and days even more so. A product dimension hierarchy
can consist of a product line level, product type level, and a products level.
It is possible to have multiple hierarchies in a dimension. For example, a time hierarchy can
describe the time elements in a calendar year. Another hierarchy might organize time in a
fiscal year. Each hierarchy is a way to classify data in an organizing structure.
Levels can consist of attributes of a dimension or expressions that operate on attributes. For
example, a store dimension might have a hierarchy that organizes stores by the attribute of
size. A level might consist of an expression that takes individual store sizes and groups stores
with similar-sized stores.