Received: by 10.68.241.7 with SMTP id we7mr6868500pbc.4.1337013185568;
        Mon, 14 May 2012 09:33:05 -0700 (PDT)
Path: pr3ni25498pbb.0!nntp.google.com!news1.google.com!postnews.google.com!ri8g2000pbc.googlegroups.com!not-for-mail
From: joel garry <joel-ga...@home.com>
Newsgroups: comp.databases.oracle.server
Subject: Re: Slower performance after enabling async io on Oracle Linux
Date: Mon, 14 May 2012 09:17:31 -0700 (PDT)
Organization: http://groups.google.com
Lines: 39
Message-ID: <01eb7374-776f-4288-8342-1ce234a5054a@ri8g2000pbc.googlegroups.com>
References: <27769382.324.1336986236654.JavaMail.geo-discussion-forums@vbyg17>
NNTP-Posting-Host: 64.58.150.28
Mime-Version: 1.0
X-Trace: posting.google.com 1337013185 17229 127.0.0.1 (14 May 2012 16:33:05 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: Mon, 14 May 2012 16:33:05 +0000 (UTC)
Complaints-To: groups-abuse@google.com
Injection-Info: ri8g2000pbc.googlegroups.com; posting-host=64.58.150.28; posting-account=tpQovAkAAABNoH5bwsZAiff2L0zxGwdv
User-Agent: G2/1.0
X-HTTP-UserAgent: Mozilla/5.0 (Windows NT 5.1; rv:12.0) Gecko/20100101 Firefox/12.0,gzip(gfe)
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On May 14, 2:03=A0am, Matthias Hoys <matthias.h...@gmail.com> wrote:
> Hello,
>
> Platform is Oracle Linux 6.2 64-bit with Oracle 11.2.0.3 64-bit.
>
> After enabling async io, I see slower performance for data pump exports, =
RMAN backups and long-running queries (between 10 and 15% slower).
>
> I enabled async io with:
>
> ALTER SYSTEM SET FILESYSTEMIO_OPTIONS=3DSETALL SCOPE=3DSPFILE;
> aio-nr is 19052 and aio-max-nr is 1048576. Storage is on iSCSI.
>
> Anyone else with the same experience? Is extra tuning for async io possib=
le on Oracle Linux?
>
> Thanks,
> Matthias Hoys

Well, it could be simply squeezing the bottleneck balloon, where
something goes faster and consequently something depending on it slows
down (for example, more cpu required to service more I/O).  Some
hugepages misconfiguration could have strange consequences simply from
wasted memory, too.

Also see, for example, http://blog.tanelpoder.com/2010/11/23/asynch-descrip=
tor-resize-wait-event-in-oracle/
fixed by your patch, but the linux/oracle async interaction is so
delicate, you may be seeing something unfixed by the patch.  Any
errors like
http://www.freelists.org/post/oracle-l/WARNINGio-submit-failed-due-to-kerne=
l-limitations-MAXAIO-for-process128-pending-aio127,2
?

jg
--
@home.com is bogus.
http://www.utsandiego.com/news/2012/may/10/hackers-taking-aim-small-medium-=
businesses/