News for hackers. Stuff that matters.
Re:Maybe Normalizing Isn't Normal by gaius on Tuesday July 15, 2008 @06:26AM (#246539)
Firstly, he has misunderstood database design. The design that the system analysts product will be in 3NF because their job is to understand the data in its most abstract sense. They will also specify - again, fully normalized - the main queries that will need to be run. Then it will go to the implementors who will transform the tables and the queries for performance. Denormalizing too early is a premature optimization. Not denormalizing at all is inexperienced.

Secondly, if you need to scale you are most likely using a query cache (e.g. Tangosol) between your RDBMS and your middle tier, so you can preserve a far cleaner data model now then you could back in the day.

Re:Maybe Normalizing Isn't Normal by biohacker42 on Tuesday July 15, 2008 @10:30AM (#246675)
I think a lot of people also completely miss the fact that there is more then one way to have a normalized DB. Much like OO design, there's a big gap between smart,good, efficient implementations and the rest.