Database Reference
In-Depth Information
2.3 Quality Data eQuals a Quality essBase DataBase
As I hope my artfully disguised, yet all too real, Essbase horror stories showed, an
Essbase database worth its salt is built on the foundation of good data; anything else is
suitable for the trash bin and should be tossed there before you end up alongside it. now
that I have scared you witless, it is time to build you back up with an approach to data
quality that ensures you never end up with a Tale of Terror of your own.
2.3.1 How Not to Solve the Problem of Bad Data
2.3.1.1 Where Is the Worst Place to Pretend That Essbase Is Really an ETL Tool? Alas, it is
included with each and every copy of Essbase and has been there since approximately
the year dot, which just goes to show that a bad idea can gain adoption through repeti-
tion despite its manifold shortcomings. Bad tool I name thee: “Essbase Load rules.”
I speak heresy given its almost universal usage. regardless, I am the Essbase Load rule
apostate and I shall explain why you should be one as well in the most restrained man-
ner I can manage.
Lest you think I have well and truly gone of the deep end of Load rule hate, I am
not suggesting that you take up manual dimension building or loading data through
free form text files in an effort to become ideologically pure. While Load rules are a
good way to load data and create dimensions, they are very much the wrong way to do
the transformations that are the “t” in true EtL. Load rules are simply a maintenance
black hole if they are filled with transformations.
2.3.1.2 Historical Context When Essbase was new, circa 1993, Essbase data-
bases tended to have ad hoc data sources and the tool itself was the ward of the
Finance department. Less than rock-solid data sources and a lack of It involvement
meant that Essbase developers had only their wits and Load rules to rely on for
data manipulation; Arbor and then hyperion charged a pretty penny for the SQL
option and few purchased it. I was not lucky enough to often come across clients who
owned it.
time has moved on, anyone still involved with Essbase is older and hopefully wiser,
the It department is now is at least a co-owner of Essbase, and yet Load rules remain
largely unchanged with the huge exception that the SQL interface is now standard and
available without additional cost.
2.3.1.3 A Tool Not Fit for Purpose I have related how I relied on Load rules on a proj-
ect and was badly burnt. much of this was my fault for not pushing back harder to the
client when inadequate metadata and data was dropped in my lap. But, I was young,
foolish, and convinced that Load rules could do anything. They could not then and
cannot now. In retrospect, it is not hard to understand why—heavy-duty EtL is not
what Load rules were designed to do. Let my transgression of relying on Load rules
be an isolated one.
2.3.1.4 The Load Rules ETL Diatribe Space considerations have precluded me from
including my well-rehearsed lecture on why Essbase Load rules are a really bad place to
do anything but the most simple forms of EtL.
I, however, have been given room to note that there are over 45 possible settings in a
given Load rule (remember that a Load rule can do both data and dimension loading)
Search WWH ::




Custom Search