Databases Reference
In-Depth Information
Preliminary Source to Target Data Map
When developing the business dimensional model, use a spreadsheet that
contains all of the data elements from the source system that were identified
during detailed data analysis. For each element that was included in the
dimensional model, the name of the dimension or fact group should be noted.
In addition, for those data elements that were determined to have no reporting
or analytical value, this should also be noted. By keeping track of where these
elements are placed in the dimensional model, this can be used as an early
source-to-target data map to help with the ETL process, which is described in
Chapter 10.
The final notes that need to be included are those about what additional
processing or transformations will need to be done during implementation to
get to the dimensional model representation. These are not the final business
rulesforprocessing,butnotesaboutthediscussionsandperhapsideas.The
final business rules will be defined later in the project.
After all of the dimensions and fact groups that were identified during
the initial brainstorming have been drafted, the first cut of the business
dimensional model is done. There are still several weeks of work ahead, but
the basics are in place and it should be clear which areas still need the most
work.
Completing or Fleshing Out the Model
Once the initial draft of the model is completed, some of the hardest work
still lies ahead. The easiest and most straightforward data is already modeled.
Now, the hardest and most complex data must be studied. The model moves
toward completion by addressing issues, making sure that source data is
represented properly, and having business group reviews.
Working Through the Issues
As you developed the initial draft of the model, each of the detailed questions
and issues were logged. Now it is time to review these. Often, what seemed
like a very difficult topic up front is actually easy to resolve. With the entire
model laid out, the placement of these other data elements is often clear. The
remaining issues tend to fall into three categories: those requiring technical
research, those requiring business research, and those that need further con-
sideration to see how they should be included in the model.
Often, nuanced knowledge of a source system database is only found with
key individuals who have been working with that data for years. Many of
the issues that need to be addressed require the insight of this experienced
systems resource. One member of the core data warehouse project team must
