Re: Index maintenance...

"Richard Foote" <richard.foote_at_tbigpond.nospam.com> schrieb im Newsbeitrag news:zo6mc.21720$TT.506_at_news-server.bigpond.net.au...
> "Volker Hetzer" <volker.hetzer_at_ieee.org> wrote in message
> news:c7abns$lm2$1_at_nntp.fujitsu-siemens.com...
> > Hi!
> > We've got a consultant (for a canned app) here who just hinted at a
> possible performance improvement
> > if one recreates the indexes after a big load.
> > He said he noticed an improvement but didn't run any benchmarks. He used
> 8.1.something and 9.0.something.
> > We are using 9.2.0.4.
> > So, what's behind this?
> > Does this apply to us too?
> >
> > Lots of Greetings!
>
> Hi Volker,
>
> When I hear someone say something like that, two questions immediately
> spring to mind.
>
> The first is simply why ? Ask the consultant why should a big load result in
> an index now being detrimental to performance.
It's not a db consultant. He just told about his experience. Furthermore, it turns out that we can't reproduce that behavior yet, simply because right now it looks like we get the performance gain (15min->20s) always, meaning, the first attempt takes long and the second not. However, we are still testing. Unfortunately we *never* get any noticeable load on the db, yet the behavior persists if we restart their server. But we're still on it.
First I'd like to abstract from db caching, then I try rebuilding the indexes manually, then we'll know whether their server just rebuilds the indexes or messes around with other things.

Lots of Greetings!
Volker Received on Wed May 05 2004 - 09:41:35 CDT