Database Reference
In-Depth Information
creating a semantic data model? Does this not appear to be a practical design
approach?
Although the approach is deceptively simple, as you will note in Chapter 10, this
method is subject to serious problems if the tables are not defined properly. You
are likely to end up with a faulty set of tables in your relational data model, with a
high potential for data corruption and inconsistency.
Dr. Codd suggested an ordered methodology for making this design approach
work. After an initial set of tables is designed, you must go through a step-by-step
process of normalization of the initial tables. After the normalization steps are com-
pleted, your relational data model will result in a set of tables that are free from
redundancies and errors.
Here are the steps in this design approach:
•
Gather the information requirements of the organization.
•
Review the information requirements to determine the types of tables that
would be needed.
•
Come up with an initial set of tables.
•
Ensure that your initial set of tables contains all the information requirements.
•
Normalize the tables with a step-by-step methodology.
•
Review the resulting set of tables, one by one, and ensure that none of the tables
has potential redundancies or errors.
•
Complete the relational data model.
•
Review the relational data model for completeness.
Chapter 10 covers this approach to designing the relational data model in detail.
You will recognize the need and motivation for the normalization process. We will
list the normalization steps and show how to apply a normalization principle at each
step. You will note how, after each step, the set of tables comes closer to being the
correct set of tables and becoming part of the final relational data model.
Evaluation of the Two Methods
Naturally, when there are two ways for arriving at the same place, you must ask
which path you should take. If both methods produce the same desired result, which
method is more appropriate? The answers to these questions depend on the cir-
cumstances of the design process.
Note the following points about the two methods while making the choice
between the two ways:
Same result.
If you carry out the transformation of the semantic data model into
a relational model or adopt the traditional method using normalization, you will
arrive at the same relational data model. However, either method must be used
carefully, making sure that every task is executed properly.
One method intuitive.
In the traditional method, you are supposed to come up with
an initial and complete set of tables. But how do you come up with the initial set?
Using what method? There is no standard method for arriving at an initial set of
Search WWH ::
Custom Search