Here's a very cursory shot at it:

Why is normalization a good thing? It saves space and eliminates anamolous data. In small tables, the space saving is not such a big deal. In tables that aren't updated, the anamolies are more easily avoided.

There can be a performance gain by avoiding joins and denormalization is often employed for exactly this reason. If your apps are doing only selects against a small table, then I'd say go with the DBA on this question.

However, if you're talking about doing inserts and updates against the table, then you have a different situation.

One approach is to keep data in normalized tables for maintenance, then do joins to create denormalized tables to run against. Sometimes the normalized tables are in a separate database, joined there, then exported.

By the way, what RDBMS are you using?

adamsj

They laughed at Joan of Arc, but she went right ahead and built it. --Gracie Allen