support data in a data base for a portion of a company. For example, a local data mart could be designed concerning employees, such as for sales personnel as-
signed to the St. Louis branch office. Detailed customer and product information would be contained in this sales employee data mart. Alternatively, a segment data mart could be constructed for a major subset of the firm, such as for sales man-
agers of the Midwest region. This segment data mart would contain detailed and summarized sales and marketing information. A data warehouse, by contrast, stores copies of decision support data in an integrated data base for an entire en-
terprise: Sometimes an enterpriseWide data warehouse, which may store hundreds of terabytes of data, is formed when- - a ' film integrates its local and segment data marts.
A.data mart or data warehouse contains tables of integrated data-base files derived from legacy or ERP systems and nonrelationa I data obtained from other diverse sources, such as market research, credit reports, or newspaper articles. As opposed to applications-oriented data in legacy systems, data in a data mart or warehouse are stored by subject areas. (e.g., customers, products, and vendors). A data mart or a data warehouse can be built with an existing microcomputer-based data-base management system software package.*
Then a portion of the legacy or ERP data, as well as data from other sources useful to decision makers, is copied to one or more data-base servers (e.g., mainframes, minis, and/or microcomputers) that are interconnected to other computers within a LAN-based client/server setup. End users' desktop computers, called clients, can readily access this data.
Since the stored data are periodically updated, the data mart or data warehouse contains "views" of recent informational data and archives of informational data for a designated past period. Data may be stored in both summarized and detail or "raw" form. By using graphical user interface (GUI) and report generator tools, such as decision support and executive information systems software, end users can easily generate daily, weekly, and monthly predefined reports. Users can employ drill down to examine summary data in finer levels of detail. For example, an executive could view a division's summary financial results on-line and move down to the detailed data to check if profits are increasing or decreasing. The executive by examining lower levels of detail could determine the region and particular retail store(s) responsible for the increased or decreased profits. Furthermore, software can be used for data mining, a technique that finds patterns and relationships among data items in the data bases. In the previous example, a user could employ statistical software to analyze what variables are responsible for the increased or decreased profits at specific retail stores. Then a model could incorporate these variables to predict future regional stores' profits.
Users can also import data into their personal spreadsheet or data base for further analysis. Decision makers can also quickly produce a variety of customized, innovative one-time reports and analyses. Finally, the data mart and data warehouse informational reporting system can be designed to automatically transmit relevant reports and analyses to designated end users.