Re: Code in the database or middle tier (the CLR controversy)

From: DA Morgan <damorgan_at_psoug.org>
Date: Sat, 04 Jun 2005 06:29:18 -0700
Message-ID: <1117891638.599116_at_yasure>


JRStern wrote:

> Don't know what this has to do with CLR controversy, but ...
>
> If indeed those six alternatives are equivalent, then #4 is certainly
> the best one to give any competent RDBMS. Except that you might try
> learning ANSI join syntax.

In an Oracle class? Why? They can be done ... but zero value in the more verbose syntax.

> You've got to be joking that #5 would produce a better execution plan
> on any real database,

You've got to be joking if you think that is what he demo proves. It does exactly the opposite. #5 is an intentional disaster area.

> Are you telling us that you saw better *peformance* between one plan
> and another

Exactly. Otherwise what would be the point in doing this demo in front of a university class?

Do keep in mind the audience for this thread is at least four separate RDBMS products. Your mileage may vary. And do keep in mind also that, as both Erland and I have both observed, different versions of the same product do different things. Adding indexes, and the type of index, etc.

-- 
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)
Received on Sat Jun 04 2005 - 15:29:18 CEST

Original text of this message