Databases Reference
In-Depth Information
Gaining Final Commitment
The model should be presented to everyone who participated in the require-
ments gathering process. The meeting should be like the previous business
review sessions, and minimal changes should result from this session. Each
of the business people who participated in the development and refinement of
the model need to play an active role in this session. It is much more effective to
have business-to-business dialogue about the model than for systems people
to feel they need to defend their work. If changes need to be made, this is still
a good time to do so.
If the model does indeed represent the business perspective, the audience
will be pleased, but perhaps not openly thrilled. This should be viewed as
success! It means that the model makes sense to them, which is the goal.
Expanding Business Data Over Time
Dimensional models do indeed change and expand over time. This happens
as a natural course of business. There are two common ways for dimensional
models to expand: enhancing dimensions and adding new facts.
Enhancing Dimensions
As more, newer, or cleaner sources become available, new attributes can be
added to existing dimensions. As long as these new attributes do not change
the grain of that dimension, there is no impact on existing fact groups. The
business community gains the ability to create new reports using any of the
existing facts.
Dimensions can be fairly easily rebuilt, as long as the dimension does not
have 50 million instances. If the dimension is extremely large, this can still be
done, but it requires some additional planning and time. Businesses that deal
with these extreme volumes are well versed in dealing with this much data, as
it pervades all systems development, not just data warehousing.
Adding More Fact Groups
Another common way that dimensional models expand is with the addition
of new fact groups. Usually this happens when new facts, often from new
data sources, are loaded. These new facts must be studied to determine their
dimensionality and grain. The new facts may be added to existing fact groups
as long as they have the same dimensionality and grain, although the technical
team may decide to implement these separately. New fact groups can be added
without impacting or changing any of the existing fact groups. Existing users
of the data warehouse may not see any change. The new data can be released
to only those who need access to it. Of course, all dimensions used by the
existing and new fact groups must be conformed.
