Re: Q: Perfromance problems on AIX

From: <cwalkersddpc_at_my-deja.com>
Date: Wed, 12 Jan 2000 20:18:56 GMT
Message-ID: <85inj0$i0b$1@nnrp1.deja.com>

Check the ULIMIT for the UNIX user that is running the UNIX process that is running statement.
This will usually be similiar to this:

oracle 13252 1 0 06:41:38 ? 0:14 oraclesupport (LOCAL=NO)

In this case the unix user Oracle is running the process. Log in as that user and run this statement at the UNIX prompt: ulimit -a

The output will be similiar to the following:

time(seconds)        unlimited
file(blocks)         2097151
data(kbytes)         131072
stack(kbytes)        32768
memory(kbytes)       32768
coredump(blocks)     2048

nofiles(descriptors) unlimited
Check the output on both machines and compare them. If they are different make them the same and try again.

In article <387C9981.BA838DE7_at_quantum.de>,   Waldemar Schlonsok <Waldemar.Schlonsok_at_quantum.de> wrote:
> Hallo again,
>
> jobrien99_at_my-deja.com schrieb:
>
> > Also run Analyze on the tables.
>
> Do you mean, to use a cost based optimizer? I think, i should first
try to find
> the leak either in the base or with the server.
>
> > Run sar 5 100 and observe the wio on both machines.
>
> OK. But what it means there is a heavy WIO on the system. I does not
have any
> experiences with tuning OS or either
> analyzing performances of OS. Usually a high number means a problem,
but how
> high ist normal?
>
> > In article <85dglb$noa$1_at_nnrp1.deja.com>,
> > michael_bialik_at_my-deja.com wrote:
> > > Hi.
> > >
> > > Try comparing EXPLAIN on test vs. prod machines.
> > > If no differences found -
> > > use ALTER SESSION SET SQL_TRACE = TRUE;
> > > to trace what actually happens in prod. ( May be you are
> > > using incorrect access and have very small number of rows in test
).
>
> I does not like to tune the application in this situation. I have a
problem with
>
> different machines, the nearly the same data bases on both of ist. And
the
> cruix is, the application on the slower machine is much faster (up to
two times
> faster)
> then on the faster machine.
>
> Thanks for hints.
>
> Yours sincerely,
> Waldemar Schlonsok
>
>

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Wed Jan 12 2000 - 14:18:56 CST