Re: Optimisation of a sql query

From: joel garry <joel-garry_at_home.com>
Date: 21 Jul 2006 15:50:49 -0700
Message-ID: <1153522249.228615.103210@m73g2000cwd.googlegroups.com>

Bob Jones wrote:
> "joel garry" <joel-garry_at_home.com> wrote in message
> news:1153404182.225435.310230_at_i3g2000cwc.googlegroups.com...
> >
> > Bob Jones wrote:
> >> > And Mladen is wrong about one thing, portable SQL will not suck equally
> >> > on every database. It will likely suck extremely badly on some
> >> > databases, but only suck very badly on others.
> >> >
> >>
> >> Non-portable SQL has no chance to suck on other databases at all.
> >
> > And that's a bad thing?
> >

> That is a very bad thing when one has to move SQL from one database platform
> to another.

>


> > Bob, I work daily with portable code.  I deal with unnecessary problems

> > daily with portable code.  In fact, it is quite lucrative, as few

> > people understand both the semantics of the code (it's not SQL, but

> > rather a 4GL that generates SQL for OCI) and the issues on the Oracle

> > side, and how to deal the frustrations involved in not being able to

> > fully utilize Oracle.  It actually does perform amazingly well, if you

> > ignore the lockups due to a non-Oracle locking and concurrency model,

> > and having to manually fix flags in the data when things go south, and

> > ignore some pretty poor programming by high priced consultants.  I

> > guess you could say that about any cross-db enterprise software, eh?

> > At least this one gives customers all the source - which means any site

> > will likely have some verrrrrrrry strange code.
> >

>
> We are only talking about SQL, are we not?

Certainly. Did you miss the part about OCI? Did you miss the part about different locking/concurrency models _requiring_ different SQL for good performance on each?

That's the intractable problem with portable SQL: the different models require different application design. Tom Kyte does an excellent job demonstrating the details of this in a few chapters of a few books, but that's the issue in a nutshell. Now, it's possible that newer features of databases may allow this problem to go away, but I haven't seen any enterprise class application software that demonstrates that yet. Moving database features from the db to another tier not only doesn't count, it is a negative.

--
@home.com is bogus.
What do you mean, we're out of toner?  I need a new kidney!
http://www.signonsandiego.com/uniontrib/20060720/news_1c20bioi-jmp.html

Received on Fri Jul 21 2006 - 17:50:49 CDT