From a normalized model, dimensional data marts, views or aggregates can be created to support specific sets of requirements, while maintaining overall consistency and without skewing data organization solely around needs of a subset of particularly vocal or politically well positioned users.

WhereScape RED is particularly adept at building out a dimensional or presentation layer from an existing data warehouse.

A “single version of the truth” can be achieved through a consistent model feeding downstream views or aggregations. This is a requirement for large scale enterprise data warehousing where data volumes and the need to cater for constant change dictates a modeled area independent of source systems and specific, changing, requirements.

Data for a normalized model is commonly passed through a series of staging environments where it is transformed from source system specific models (optimized for individual applications rather than business consistency) into a business specific model. WhereScape RED supports both transient and permanent staging tables, the decision on which to use should depend on volumes and the ability to recreate extracts.

While a normalized model future proofs an enterprise data warehouse, the majority of users (and query tools) prefer to interact with denormalized or multidimensional structures. This is not in conflict with normalized designs. A multidimensional view of data is common in an enterprise data warehouse, and can be implemented as a dependent data mart, a view, an aggregate table or a multidimensional cube. Combining a dimensional presentation layer with a normalized data warehouse model creates a powerful solution for enterprise data warehouse environments.