Clear all

Hyper generalized data models

Posts: 19
Topic starter
Active Member
Joined: 4 months ago

While Ralph Hughes has a chapter (Chapter 14) on the hyper normalized form (HNF) approach (such as DV2) to the EDW in his book Agile Data Warehousing for the Enterprise, he follows it with Chapter 15 titled "Fully Agile EDW with Hyper Generalization".

He writes:

"For other EDW teams, however, the thought of employing a hyper normalized form only creates concern. These teams are disquieted by the fact that HNF addresses only the integration layer of the data warehouse, that it doubles the number of tables of a standard normal form model, and that it makes data considerably more difficult to retrieve. They see the performance workarounds such as seeding null records and materializing point-in-time tables as kluges and wish there was a more widely supported means of generating a far more complete enterprise data warehouse.

A more complete solution does exist, and it lies in the opposite direction along the hyper modeling spectrum. Instead of hyper normalizing an EDW model into more tables, the development team should hyper generalize the integration layer down to just a handful of entities."

Has anyone explored hyper generalization?

1 Reply