Databases Reference
In-Depth Information
deliver the Prior Day Sales report and a dozen other reports that are only
slight variations of the Prior Day Sales report—for example, the Prior Day
Department Sales report, the Prior Day Sub-Department Sales report, and
the Prior Week Sales report. What about the Prior Day Logistics report?
hat's another subject area and another purpose. Remember, one subject
area at a time.
relational Integrity
Another good clue that a data warehouse is not ready for prime time is
the multiplication and disappearance of data. When the Prior Day Sales
report indicates that sales yesterday were $500,000, but the Prior Day
Department Sales report indicates sales yesterday were $750,000, some-
thing is amiss in the data warehouse. Or, when the Prior Day Department
Sales report indicates the Furniture department disappeared, something
is not quite right in the data warehouse.
Again, this would seem to be rather obvious. That is why it is a good
first indicator of less than stellar success. The dimension tables look like
dimension tables. The dimension and fact tables join like dimension and
fact tables. As individual elements of a data warehouse they all seem to
fit the descriptions of a data warehouse. But, when joined together in the
form of a report or query, they distort the data returned by the data ware-
house. Why is that? Again, from the first planning meeting to the final
report review the fact tables and dimension tables will be designed to cor-
rectly join to achieve the result set for a Prior Day Sales report if the Prior
Day Sales report is the goal of that subject area. By using the Prior Day
Sales report to guide the relational integrity of the Sales subject area, other
reports and queries will also experience good relational integrity. In short,
the relational integrity necessary to make the Prior Day Sales report a suc-
cess will still be there for other reports. By achieving its purpose, which
is the Prior Day Sales report, a data warehouse experiences the secondary
success of achieving other ancillary purposes as well.
Data Quality
The third key indicator of the success or failure of a data warehouse is not
quite so obvious. However, what it lacks in obviousness, it compensates for
in simplicity. Why do less successful data warehouses not include a Data
Quality program? The answer is simple: they don't know what bad-quality
