Databases Reference
In-Depth Information
data looks like. A word looks exactly like a word; a non sequitur word
looks exactly like every other word. But, when the only acceptable values
for a data element are
north
,
south
,
east
, and
west
a data quality application
can rather easily identify
yes
as a non sequitur word.
A Data Quality program is diἀ cult. You can take that plain and simple
statement to the bank. In a Data Quality program you programmatically
lock your definitions of acceptable data values and bad data values in logic
written into ETL code. To do so, you must know what acceptable and bad
data look like, and you must be able to define them in terms of program-
matic logic. That task, all by itself, is diἀ cult. Typically, a Data Quality
program is an iterative effort. Each iteration is a refinement of prior efforts.
So, the mere presence of an active and functioning Data Quality pro-
gram is a key indicator of a successful or failing data warehouse. It's that
simple. The presence of an active and functioning Data Quality program
means that someone is validating some of the data in the data warehouse
to verify that it is consistent with the logic and design of the data ware-
house, which is much better than no one validating any of the data in a
data warehouse.
How is a Data Quality program enhanced by the presence of a purpose?
The purpose of a d ata warehouse provides a focus to the Data Quality
program. If the purpose of a data warehouse is to produce a Prior Day
Sales report, then the Data Quality program will begin by validating the
data that contributes to the Prior Day Sales report. Even if the effort is
incomplete, yet iterative, any attention to data validation is better than no
attention to data validation; furthermore, validating the data delivered to
the members of management who choose to fund the data warehouse is
always a good choice.
PurPose
If you choose to aim at nothing, you just may hit the target.
A data warehouse, like every other enterprise asset, must have a reason
for existing. If a data warehouse is not busy adding value as a data ware-
house, it may soon begin adding value as something else. While ROI is
thought to mean Return on Investment, for those operating a data ware-
house ROI means Reason for Existing. The value added by a data warehouse
contributes to the ability of a data warehouse to maintain its integrity in
