[
Lists Home |
Date Index |
Thread Index
]
with all due respect len, this argument was dealt with comprehensively
about 30 years ago. databases that aren't fully normalised will always
be more efficient in particular cases, but not in the general case.
moreover there's no mathematical proof of correctness when things aren't
nomalised, whereas there is when they are.
however there are considerable performance issues surrounding the
optimisers in many of the popular rdbms's and their sql implemenations.
nevertheless, the case in point here is how to support two simultaneous
tree views. xml supports one easily, as in eg orders and the lines on
the orders. but what if you want a second that is products and the
orders they are on. one is quick and efficient, the other is very slow.
but we discussed this in detail a long time ago.
rick
On Wed, 2003-06-18 at 23:12, Bullard, Claude L (Len) wrote:
> Keep in mind that 'bad normalization' can equal
> 'good optimization'. I seldom see fully-normalized
> databases because the overhead of getting and
> opening tables, creating the view, etc. can be
> prohibitive vs the maintenance of denormalized
> data. Otherwise, I would agree. These aren't
> legacy databases. They are practical ones where
> performance considerations dominate maintenance.
>
> len
>
> From: Joe English [mailto:jenglish@flightlab.com]
>
> What are the real-world use cases of minOccurs and maxOccurs?
> In my experience, occurrence constraints specifying anything
> other than zero, one, or many are almost always an indication
> of a bad design decision somewhere in the system.
>
> (Or maybe that *is* the use case? To be able to accurately
> describe badly normalized legacy RDBMSs and other such things?)
>
> -----------------------------------------------------------------
> The xml-dev list is sponsored by XML.org <http://www.xml.org>, an
> initiative of OASIS <http://www.oasis-open.org>
>
> The list archives are at http://lists.xml.org/archives/xml-dev/
>
> To subscribe or unsubscribe from this list use the subscription
> manager: <http://lists.xml.org/ob/adm.pl>
|
