Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


The logic at the heart of the argument for the storage of massive amounts of detail for DSS processing is hard to argue with. Intellectually, it must be correct to say that you need as much detail as possible for DSS and EIS processing. But,


in some important ways, the argument suggests Zeno’s paradox. In Zeno’s paradox, logic inescapably “proves” that a rabbit can never outrun a turtle as long as the turtle has a head start on the rabbit. Of course, reality and our own observations tell us something quite different, warning us that any conclusion based purely on logic is circumspect.


What, then, is so wrong with keeping all the detail in the world around when you are building a DSS/EIS environment? There are several things wrong. First, the amount of money required for both storage and processing costs can go sky-high. The sheer cost of storing and processing huge amounts of detailed data prohibits the establishment of an effective EIS/DSS environment. Second, massive amounts of data form an obstacle to the effective use of analysis techniques. Given very large amounts of data to be processed, important trends and patterns can hide behind the mask of endless records of detailed data. Third, with the detail, reuse of previous analysis is not fostered. As long as there is a massive amount of detail around, DSS analysts are encouraged to create new analyses from scratch. Such a practice is wasteful and potentially harmful. When new analysis is not done in quite the same way as older analysis, very similar analyses are done and ironically conflicting results are obtained.

Скачать в pdf «Building the Data Warehouse»