Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»

Example of going from data to information:

“How has account activity been different this year from each of the past five for the financial institution?


current value— 2 years

1 year

Figure 1.8 Existing applications simply do not have the historical data required to convert data into information.

curent value—

18 months

ancy between the time horizon (or parameter of time) needed for analytical processing and the available time horizon that exists in the applications.

A Change in Approach

The status quo of the naturally evolving architecture, where most shops began, simply is not robust enough to meet the future needs. What is needed is something much larger—a change in architectures. That is where the architected data warehouse comes in.

There are fundamentally two kinds of data at the heart of an “architected” environment—primitive data and derived data. Figure 1.9 shows some of the major differences between primitive and derived data.

Following are some other differences between the two.

■■ Primitive data is detailed data used to run the day-to-day operations of the company. Derived data has been summarized or otherwise calculated to meet the needs of the management of the company.


•    application oriented

•    detailed

•    accurate, as of the moment of access

•    serves the clerical community

•    can be updated

•    run repetitively

•    requirements for processing understood a priori

•    compatible with the SDLC

Скачать в pdf «Building the Data Warehouse»