Data: Everchanging and Eternally the Same - High-Performance Web Databases: Design, Development, and Deployment - page 570

data — a summarized view of the company transaction data — was a non-
atomic summary whose definition needed to be changed. Had anyone on
the project team clearly understood this situation at the time, the results
might have been more pleasant. The actual results were that a massive
overtime effort was undertaken to (1) re-stating the current transaction
data and (2) approximate the historical data. There were two major gaps:
(1) the historical data was actually re-stated a year later, so the “badness”
of the approximation was then easily measured; and (2) since no one at the
time clearly understood the impact of the approximations, no one could
clearly communicate to the executive how good or bad the approximation
was. Common sense caught a lot of bad approximations as checks were
about to go out the door, but the cost was still considerable.

This anecdote points out two realities: (1) even though the categoriza-
tion of data mismatches seems simple on first reading, it can be quite
complex when applied to a real world situation; and (2) no matter how
clever one is, it is easy to lose sight of the swamp when the alligators are
closing in.

BUSINESS RULE MISMATCHES

The bigger the organization, the more energy it will take to get these
sorted out. The more history an organization has, the worse the business
rule matches are likely to be.

To read the marketing brochures of information warehouse providers is
to wax rhapsodic in a reverie of total information availability — every fact
and facet of the business instantly available at executive fingertips, all re-
lationships neatly sorted, and all data readily comparable with other data.
So if the technology is here, why isn't the dream a reality?

. Remember that there are fairly
explicit directly data related business rules like cardinality and referential
integrity. There are also all those “other” nuances of data interpretation
that are imbedded in all those queries, spreadsheets and programs that are
lurking out there (see Exhibit 39-2).

My answer: business rule mismatches

Here is a good example that will cause lots of telco problems in the next
few years. There is a historical association between NPA-NXX (known to
most people as “area code” and “exchange”) and a geographic location.
The NXX normally corresponded with a physical structure called a frame.
In fact, in the past, if you knew someone's NXX (exchange) in an urban area
you could tell the general neighborhood they lived in. In the thousands
upon thousands of programs that have been written to handle Telco billing
and provisioning, the assumption about that connection has been so deep-
ly embedded as to be hardly detectable. The same assumptions have been
embedded in hundreds of file and database designs. This is still important