Databases Reference
In-Depth Information
Reflecting on Business Realities: Advanced
Concepts
For many people, the basic dimensional modeling concepts presented so
far will be sufficient for their needs. For those who want to learn about
dimensional modeling in more detail or who are about to participate in the
modeling process, several additional topics are included here. This is not
intended to make you a dimensional modeler, but it describes techniques that
you can use to address many common, complicated situations. The business
dimensional model notation to identify these situations is presented. This
notation provides the technical team with an indication that a condition exists.
Standard dimensional modeling techniques can then be used to complete the
actual database design.
Supporting Multiple Perspectives: Multiple Hierarchies
It is a common misconception that all groups within the organization must
agree on the structure and content of one hierarchy. Not only is this not
true, but multiple hierarchies are encouraged because this enables multiple
viewpoints to be represented in the model. In the Date dimension presented
earlier (refer to Figure 7-3), note that there are two hierarchies: one to support
thecalendaryearrollupandonetosupportthefiscalyearrollup.
Another example of the benefits of multiple perspectives can be seen for
customers. In this example, the ''ship to'' location is what is called Customer.
Figure 7-15 shows multiple customer hierarchies: the customer's location, the
customer's bill to location and the company's sales organization hierarchies.
Each hierarchy is diagrammed separately in the business dimensional model
so that each is clearly visible. It is important to ensure that each is labeled
uniquely.
Tracking Changes in the Dimension: Slowly Changing
Dimensions
Most of the data in the data warehouse that is tracked over time are facts.
This is a natural and basic part of dimensional modeling. However, there will
be changes to values in dimensions that need to be addressed. There are also
cases where the dimensional attributes must be captured at the time that a
business event occurred. There are technical techniques to address tracking
these dimension changes over time.
From a business perspective, you need to evaluate which changes must
be recorded and kept. If a person's name changes, is that important for
analytical results? Perhaps not. In many cases, the current value of an attribute
