The Kimball Group has been involved in data warehousing and business intelligence (BI) for more than two decades. Dimensional modeling helps us consistently give business users the usability and performance they need. But misconceptions about dimensional modeling often trip people up. One school of thought argues that the detailed data in a data warehouse must be stored in third normal form because it's more flexible than a dimensional model. Not so! Third normal form was invented to support fast, concurrent, interactive updates of attribute values—something you rarely do in a data warehouse or BI system. Fully normalized models support transaction processing, and dimensional models support business decision making. Different problems require different solutions.
A second misconception is that dimensional models contain only summary data. Again, not so! Relational dimensional models and their corresponding Analysis Services databases should include the most detailed data that's available.