Denormalization, Data Marts, and Data Warehouses
The following quotation is taken from the web site of Bill Inmon, who coined the term data warehousing. It supports the position argued here: the more general the analyses undertaken on the warehouse data store, the more important the requirement that the data be normalized. The audience size issue he raises is a reflection of the diversity of analysis anticipated and the need to support any and all potential explorations of the data.
“The generic data model represents a logical structuring of data. Depending on whether the modeler is building the model for a data mart or a data warehouse the data modeler will wish to engage in some degree of denormalization. Denormalization of the logical data model serves the purpose of making the data more efficient to access. In the case of a data mart, a high degree of denormalization can be practiced. In the case of a data warehouse a low degree of denormalization is in order.
“The degree of denormalization that is applicable is a function of how many people are being served. The smaller the audience being served, the greater the degree of denormalization. The larger the audience being served, the lower the degree of denormalization.”