2.2. What a Data Warehouse Is
Bill Inmon is an American computer scientist and is recognized by many as the father of the data warehouse concept. His definition of a data warehouse is widely accepted.2 A data warehouse is a subject-oriented, integrated, time-variant, nonvolatile collection of data in support of management’s decisions.
“Subject oriented” means that the collected data are related to a specific business area—for example, a sales and marketing department. However, in practice, the data warehouse can include different business areas. The data warehouse is restricted to one business area or even to one department, usually called a data mart, which is a local warehouse that is generally a part of a corporate warehouse.
“Integrated” refers to the fact that all data are homogenous in their format, there is a range of acceptable values, and data are provided from most or all of an organization’s operational system.
“Time variant” means that all data entered into a data warehouse have a timestamp, which allows companies to track changes and perform analytical research at certain time intervals.
“Nonvolatile” signifies that data, once entered into a warehouse, are remembered and cannot be deleted. After downloading, one can only read the data.
Technologically, data warehouses are operated through database management systems; usually these are databases using a relational model.3 Inmon amended his classic definition of the data warehouse to include a corporate information factory (CIF).4 This is a data warehouse that contains all the information resources of an enterprise and the data coming from the external world.