Re: [OT] HOWTO's of messing up Linux cluster?

Right. It probably isn't Linux at all, but a poorly designed application that can't work on RAC. I once saw an application that got worse performance when moved to RAC. The cache fusion waits were the big problem because they would do things like, insert a row, immediately select the row, update the row that was inserted, then query it again. This was very high volume and with no node affinity. The data blocks would thrash back and forth between the 2 nodes.

Mike

>From: David Sharples <davidsharples_at_gmail.com>
>Reply-To: davidsharples_at_gmail.com
>To: oracle-l_at_freelists.org
>Subject: Re: [OT] HOWTO's of messing up Linux cluster?
>Date: Sun, 21 Nov 2004 19:34:13 +0000
>
>or a rubbish appliction that isnt scalable
>
>
>On Sun, 21 Nov 2004 14:16:36 -0500, Paul Drake <bdbafh_at_gmail.com> wrote:
> > I should have more thoroughly qualified that with "not a
> > recommendation, but you might find a setting or 2 at ..."
> >
> > start with the puschitz site and howard rogers site for oracle linux,
> > and the platform-specific docs probably wouldn't hurt either.
> >
> > Paul
> >
> > On Sun, 21 Nov 2004 11:10:49 +0100, Alexander Gorbachev
>--
>http://www.freelists.org/webpage/oracle-l

--
http://www.freelists.org/webpage/oracle-l