Design For Decision Support and Tactical Analysis
Not only is basic schema normalization an important design consideration, but normalization tailored specifically for the analytical needs of business intelligence is also critical. Even fully normalized data warehouses are often built as if they were designed to support OLTP applications.
Designers often fail to examine the following list of considerations:
These factors all support the immediate and long-term benefits provided by a fully normalized enterprise data model. The fully normalized schema provides a framework both for continued growth and for increasing user insight.
The moral of the story is this: it is often a mistake to compromise your logical data model when you select tools to extend the value of your data and to optimize the delivery of information to your users. When necessary to facilitate ease of use or to improve tool performance, implement views to help ensure the continued integrity of your logical data model. When you are evaluating the data model you need to support your data warehouse, ensure that model not only supports your business today, but that it is fully capable of supporting the enterprise well into the future.