Re: Nearest Common Ancestor Report (XDb1's $1000 Challenge)
Date: Tue, 01 Jun 2004 21:23:33 -0300
Message-ID: <pan.2004.06.02.00.23.33.211065_at_dutra.fastmail.fm>
Em Wed, 02 Jun 2004 00:17:08 +0200, Hugo Kornelis escreveu:
> Denormalizing to increase performance. I'm sure you've heard about it
> before.
That is for SQL, not for the relational model.
As you sure already read in Date or Pascal, unlike SQL the relational model does not condition univocally the physical representation to the logical model; therefore one can optimise the physical database as much as one wants without denormalising.
That said, it is *the* part of the relational model that has never been properly implemented... don't know about IBM BS12, but Oracle's CLUSTERs don't always perform as advertised, and Dataphor hasn't yet dealt with the physical storage backend, using SQL DBMSs instead. The TransRelational people have yet to forfeit secrecy and deliver something, so that perhaps the first here will be our own Alfredo Novoa.
To be strict, there is no such a thing as denormalisation... normalisation is a logical concept, while performance improvements are physical. Too bad SQL confuses both.
-- Leandro Guimarães Faria Corsetti Dutra +55 (11) 5685 2219 Av Sgto Geraldo Santana, 1100 6/71 leandro_at_dutra.fastmail.fm 04.674-000 São Paulo, SP BRASIL http://br.geocities.com./lgcdutra/Received on Wed Jun 02 2004 - 02:23:33 CEST