Professional SQL Server™ Analysis Services 2005 with MDX

1.1. A Closer Look at Data Warehousing

In the book Building the Data Warehouse, Bill Inmon described the data warehouse as "a subject oriented, integrated, non-volatile, and time variant collection of data in support of management's decisions." According to Inmon, the subject orientation of a data warehouse differs from the operational orientation seen in On-Line Transaction Processing (OLTP) systems; so a subject seen in a data warehouse might relate to customers, whereas an operation in an OLTP system might relate to a specific application like sales processing and all that goes with it.

The word integrated means that throughout the enterprise, data points should be defined consistently or there should be some integration methodology to force consistency at the data warehouse level. One example would be how to represent the entity Microsoft. If Microsoft were represented in different databases as MSFT, MS, Microsoft, and MSoft, it would be difficult to meaningfully merge these in a data warehouse. The best-case solution is to have all databases in the enterprise refer to Microsoft as, say, MSFT, thereby making the merger of this data seamless. A less desirable, but equally workable, solution is to force all the variants into one during the process of moving data from the operational system to the data warehouse.

A data warehouse is referred to as non-volatile since it differs from operational systems, which are often transactional in nature and updated regularly. The data warehouse ...