Modeling a data warehouse

• Published in: InformationWeek (Manhasset, N.Y.) no. 564; p. 60
• Main Author: Raden, Neil
• Format: Magazine Article
• Language: English
• Published: Manhasset UBM LLC 29.01.1996; Informa
• Subjects: Computer programming; Data modeling software; Data warehouses; Data Warehousing Data Mining; Database design; Database management systems; DBMS software; Design and construction; Modeling; New technique; Product introduction; Programming tutorial; Systems design; Systems development; United States > US
• ISSN: 8750-6874

Organizations need a way to convert their data into actionable information, and one technique for accomplishing this is multidimensional modeling (MDM). MDM is a way to conceptualize business models as a set of measures described by ordinary aspects of business; it is particularly useful for sifting, summarizing, and arranging data for analysis. OLTP systems are based on entities, relationships, functional decomposition, and state transition analysis, but MDM for data warehousing is based on facts, dimensions, hierarchies, and sparsity. A fully normalized OLTP design for an order-entry system could have hundreds of tables, but MDMs are much simpler; a good MDM can be implemented in a relational database, a multidimensional database, or an object-oriented database. Aspects to consider when building an MDM are examined, and how MDMs differ from relational models is described.