Re: 3GB RAM usage by Oracle

Sybrand Bakker <sybrandb_at_hccnet.nl> wrote in message

> If you want me to name US vendors selling products where the SQL is a
> piece of *shit*, I have no problem in doing that.

Yes, bad code does not stop at any border, sorry.

> And no, adding extra indexes is not going to work, NOR is throwing
> extra hardware at the same problem.

Nonsense. You have never felt the exhilaration of adding a single function-based index and seeing the execution speed cut-in-half for 500 SQL statements. It's a rush.

> Your 'tuning' strategy (which is actually no tuning at all, but a
> *religion* you should use silver bullets as paneceas to fight
> symptoms) will actually make sure the US goof balls will *never* learn
> how to do it properly.

A religion, eh? I like that! We could call it the "Cult of Common Sense".

Any reputable Oracle professional is going to optimize the database as-a-whole (by adjusting parameters, adding indexes, &c) before tuning individual SQL statements. It's Software Engineering 101 here, and all the kids learn it in College . . . .

> The 'throwing hardware at the problem' is also a typical US stupid
> strategy, here in Europe one usually doesn't want to do that, because
> Europeans know it doesn't solve anything.

Not true. I worked with some folks in Waldorf Germany (you can guess the company) where we upgraded a CPU-bound database (already fully SQL tuned) to faster 64-bit processors. Batch jobs went from 2 hours to less than 15 minutes and everyone was delighted.

Even a properly-tuned system has a bottleneck (top-5 timed events), and the ONLY way to speed-up a well-tuned Oracle database is to hit the bottleneck.

No rocket science, Sybrand, just common sense. . . Received on Sat Aug 28 2004 - 05:43:25 CDT