 |
 |
Archives of the TeradataForum
Message Posted: Mon, 11 Nov 2002 @ 15:10:36 GMT
| Subj: | | Re: 3NF vs Dimensional modelling |
| |
| From: | | Geoffrey Rommel |
Neil Raden wrote:
| | normalization was a technique to get relational data bases to perform as OLTP platforms, because they were too slow. When Codd
devised what we now call normalization, he didn't have a word for it. | |
| | It was 1973. And you know what he said? He said, "Nixon is normalizing relations with China, so I guess that's what I'm doing,
normalizing my relations." | |
| | So there you have it, the NORMAL form is an expression that was totally arbitrary. He could have called it Third Chickenfat
Form. | |
I beg to differ. Regardless of the origin of the word, there is something "normative" about [345]NF relations: they are explicitly
designed to avoid having to update two or more tables when you are updating only one piece of data, and to avoid the well-known insert,
update, and delete anomalies. These anomalies are generally considered to be Bad Things. Moreover, the higher normal forms are not just
theoretically pure but have proved useful in real-world applications.
--wgr
| |
