Databases Reference
In-Depth Information
This is often difficult for people to understand, and it creates concern
because the reports that have been relied upon for years have been exposed
as no longer accurate. In most cases, the reports were accurate and served
the purpose as defined when they were developed. It is time to move on to
how the business is being tracked and measured today. For most people, it is
hard to give up the current way of doing things. It is comfortable; and while
not optimal, it is well understood. The changes to business rules should be
driven from business requirements and decisions made during design and
development of the ETL system. It is helpful to have the business people
who were involved in making these decisions communicate what has changed
and the rationale behind the changes. Moving forward requires learning and
changing — two things that many of us don't enjoy.
TIP
Don't focus on why the old system was wrong, but emphasize how much
better it is to use new measurements, formulas, and ways of looking at the business.
This new perspective has evolved as a result of what was learned in the past.
You are leveraging what was done in the past to make things better for the future.
Working Toward Long-Term Solutions
Depending upon what you find, some issues may simply be too big to address
at this time. If a problem originates in a major application system, the solution
is likely to be much bigger than the data warehouse. Often the team that is
responsible for the application system heartily agrees that there is a problem
and they often agree that the problem must be addressed. However, their
priorities may differ from those of the data warehouse team. The effort to
address the underlying problem may be slated with the next maintenance
release, which is scheduled to start 24 months out. The data warehouse may
not be able to wait that long. Therefore, many techniques are employed and
a lot of work is done by the ETL system to overcome these problems with
underlying systems. Over the years, real change is so infrequent that too
often data warehouse teams don't even think to ask for modifications and
enhancements to the underlying source systems.
Don't live with the status quo. Regardless of how clever the data warehouse
team is, it will be better for the long term to have clean data captured as
close as possible to the business interaction or transaction itself. The following
guidelines should help:
What is the real source of this data problem and where should it be fixed?
Encourage the data warehouse team to look for long-term solutions,
rather than short-term patches.
Follow up to ensure that modifications and enhancements are actually
made to the source application systems.
