DA Morgan <damorgan_at_x.washington.edu> wrote in message news:<1099970637.926910_at_yasure>...
> One solution is to use statistics and jobs to simulate the production
> environment. Nothing's perfect but I'll take a good case study over a
> gut instinct every time.
That's not easy, definitely not easy. Right now, we're encountering
here quite nasty problem. Our branch databases (8.1.7) are freezing
when central database (9.2.0.5) is down. We've got errors, traces,
just everything, but neither we nor Oracle support are able to
reproduce this. We're trying hard in the deployment environment,
simulating load, using the same setup, the same kind of SQLs executed
as distributed transactions, simply no way. However, our production
environment can reproduce this easily (and that's bad).
I have also a lot of experiences from many performance tests in the
past and I learned that reproducing a production load could be tough
in some cases. So I can imagine a real world behaviour that could be
almost impossible to proof using any sensible testcase and still be
perfectly valid. However, without a solid scientific evidence is hard
to make (and defend) any claims. And I also believe (maybe I'm wrong)
that most of claims unbacked by evidence made even by world known
experts should be blamed on their laziness and not on a substantial
impossibility to proof.
--
Dusan Bolek
http://www.db-support.com
Email: spambin_at_seznam.cz
Pls add "Not Guilty" to the subject, otherwise your email will face an
unpleasant end as SPAM.
Received on Tue Nov 09 2004 - 07:36:03 CST
