Database Reference
In-Depth Information
let down a project sponsor, but consider about how this bad data will impact you
direct ly.
Do you know what happens to developers who have no credibility because they
built bad Essbase databases? The most common euphemism is: “Spending more time
with his family,” aka “you're fired.” Bad data results in unpleasant consequences for
all involved and so from a self-preservation perspective alone,
bad data must be
slain
.
2.2.7 Three Tales of Data Error, Terror, and Learning
Presented for your rueful consideration are three semigruesome, real-world, as I saw
or heard them examples of data errors causing deep financial pain. These stories
are completely true yet are obfuscated and declarified enough so that the guilty can
escape the shame and contempt that is surely their due and, of course, so that “the
author” does not get sued. read on to learn the lesson that lies within each geek
tragedy.
2.2.7.1 The Error of Complexity
A customer (of someone, somewhere) had a SQL-based
budgeting system. Said system was the personification of scope creep—it did many,
many things, some of which were clearly defined. The options, code blocks, and launch-
ing processes were so multifarious and complicated that aggregate data values were
inconsistent, i.e., the system never reported the same number twice.
I was summoned and analyzed this system with an eye toward replacing it with
an Essbase database and found over 800 (yes, really) separate code streams and over
300 tables (yes, really, again). given the complexity of the system, I wondered how a cor-
rect number ever came out. It probably never did.
What was the eventual outcome (the Stones of Data Justice ground slowly, but exceed-
ingly finely) of this overly complex financial reporting system? The outcome was termi-
nation with cause for the employee who built it. never let it be said that poor data quality
cannot directly impact the size of a developer's paycheck. And what was the eventual
outcome for me and my paycheck/fee? Joy, happiness, and significant billable hours as
I fully implemented a reporting system in Essbase.
2.2.7.2 The Error of Hubris
This next story concerns a consultant, one that the author
is intimately acquainted with, and one who is willing to publicize his stupidity so that
others may not share his fate. Who is this dope? yup, it is I. At least it was a long time
ago, so memory has softened the edges of my failure.
The client was new to Essbase with Finance absolutely sold on the power of Essbase
and It much more in the wait-and-see camp. Particularly when I rather publicly proved
that a $25,000 Windows
®
server performed Essbase calculations 11% faster than the
$250,000 unix™ box It had specified. Proving that the It department was the naked
Emperor might not have been the best approach when it came to asking for their help in
data sourcing. Sometimes I only learn through pain.
I Am Not an Infrastructure Consultant
Now I am not saying Essbase runs better on Windows than on Unix. In fact, I have since seen and learned much about
Unix installations. I am just saying that the way this particular client had this particular Unix server specified and config-
ured was suboptimal. Perhaps it was too complex for them. But, this is all another matter and not germane to my story,
I just want to make sure you did not think I was dumping on Unix and you Unix lovers can tamp down your indignation.
Okay, back to the tale of your intrepid Not An Infrastructure Consultant.
