From oracle-l-bounce@freelists.org Sat Sep 10 08:09:36 2005 Return-Path: Received: from air891.startdedicated.com (root@localhost) by orafaq.com (8.12.10/8.12.10) with ESMTP id j8AD9ZgD006329 for ; Sat, 10 Sep 2005 08:09:35 -0500 X-ClientAddr: 206.53.239.180 Received: from turing.freelists.org (freelists-180.iquest.net [206.53.239.180]) by air891.startdedicated.com (8.12.10/8.12.10) with ESMTP id j8AD9ShJ006310 for ; Sat, 10 Sep 2005 08:09:28 -0500 Received: from localhost (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id B8C6E1EB070; Sat, 10 Sep 2005 08:09:22 -0500 (EST) Received: from turing.freelists.org ([127.0.0.1]) by localhost (turing [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 12494-05; Sat, 10 Sep 2005 08:09:22 -0500 (EST) Received: from turing (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 36A3C1EB00D; Sat, 10 Sep 2005 08:09:22 -0500 (EST) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:references; b=Y4BJ5Sw90ZZ93iRo3SUk7Qavcok0Y7AXlnJ93qeTO3JGpZcpbjM4BV4FlkjHcBqKIJs1JsFJkoleLVxUHMwaks+fSvGFPc2oVbxqxCjOyUXnLNqkTRsjQELq1o/cNmAgfzi+HpBiL0giKSCAgoGt4mKehP65gOas1VmIJGPDwC0= Message-ID: <367369f1050910060761412a26@mail.gmail.com> Date: Sat, 10 Sep 2005 09:07:32 -0400 From: Ranko Mosic To: Mladen Gogala Subject: Re: Are db_writer_processes and disk_asynch_io mutually exclusive on Solaris ? Thanks, rm. Cc: anysql@gmail.com, Oracle-L In-Reply-To: <1126318323l.5412l.0l@medo.noip.com> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_1925_27403192.1126357652179" References: <367369f10509090626755994e1@mail.gmail.com> <1126308048l.4267l.0l@medo.noip.com> <1126318323l.5412l.0l@medo.noip.com> X-archive-position: 25269 X-ecartis-version: Ecartis v1.0.0 Sender: oracle-l-bounce@freelists.org Errors-To: oracle-l-bounce@freelists.org X-original-sender: ranko.mosic@gmail.com Precedence: normal Reply-To: ranko.mosic@gmail.com X-list: oracle-l X-Virus-Scanned: by amavisd-new-20030616-p9 (Debian) at avenirtech.net X-mailscan-MailScanner-Information: Please contact the ISP for more information X-mailscan-MailScanner: Found to be clean X-MailScanner-From: oracle-l-bounce@freelists.org X-Spam-Level: X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on air891.startdedicated.com X-Spam-Status: No, hits=-3.8 required=5.0 tests=AWL,BAYES_00,HTML_MESSAGE, NORMAL_HTTP_TO_IP autolearn=no version=2.63 ------=_Part_1925_27403192.1126357652179 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Quote from metalink:=20 -------------------=20 From: Oracle, Frank Marshall 13-Nov-03 21:38=20 Subject: Re : dbwn VS asy io=20 You can not use multiple DBWR's with asynch IO. You turn off=20 asynch_io and use multiple db writers ( up to 10) or you use asynch_io and= =20 one DBWR. It is an either or situation.=20 ----------=20 Also metalink note Note:97291.1 says:=20 If the application is not very write intensive (or even a DSS system) and= =20 async I/O is available, then consider a single DBWR writer process;=20 ------------=20 More, there was a bug on Solaris in the past: when both db_writer_processes= =20 and disk_asynch_io were enabled performance suffered.=20 ----=20 rm On 9/9/05, Mladen Gogala wrote:=20 >=20 >=20 > On 09/09/2005 09:41:37 PM, Lou Fangxin wrote: > > We hit some problem on solaris with multiple db writer, and finnaly set= =20 > it > > to one, then the database run well. >=20 > I'm running with 3 DBWR processes, for more then a year, with async I/O.= =20 > Slowaris 9, > Oracle 9.2.0.5 , RAC. My users are storing documents into= =20 > the database as BLOB's and > that is very write-intensive, so I need multiple writers - 1 per I/O=20 > controller. > No problems. The only thing that is incompatible with async I/O are I/O= =20 > slaves, which > are intended to simulate async I/O. That is explained in the note 97291.1= . > However, on HP-UX it is not recommended to use multiple DBWR processes an= d=20 > async I/O. > This is an old information, from the time that I was working for OXHP,=20 > which used to > have 4-way OPS (not RAC) system. I remember oracle support guy telling me= =20 > that it is > not advised to use that combo on HP-UX 11.0. It was a brand new system, w= e=20 > finished > transition from an old Siemens Reliant Unix to HP-UX on Sunday,=20 > 09/09/2001. Tomorrow, > on Monday 09/10/2001 I was very busy solving various problems, but=20 > everything was successful > in the end. I thought that I will relax a bit as of the day after. The da= y=20 > after was, of > course, Tuesday 9/11/2001 and the world as I knew it ended that day. >=20 > -- > Mladen Gogala > http://www.mgogala.com >=20 >=20 > ------=_Part_1925_27403192.1126357652179 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Quote from metalink:
-------------------
From: Oracle, Frank Marsha= ll 13-Nov-03 21:38
Subject: Re : dbwn VS asy io

You can not use= multiple DBWR's with asynch IO. You turn off
asynch_io and use multipl= e db writers ( up to 10) or you use asynch_io and one DBWR. It is an either= or situation.=20

----------
Also metalink note Note:97291.1 says:
If the app= lication is not very write intensive (or even a DSS system) and
async I= /O is available, then consider a single DBWR writer process;
----------= --=20
More, there was a bug on Solaris in the past: when both db_writer_proce= sses and disk_asynch_io were enabled performance suffered.
----
rm<= br>
On 9/9/05, M= laden Gogala <gogala@sbcglob= al.net> wrote:

On 09/09/2005 09:41:37 PM, L= ou Fangxin wrote:
> We hit some problem on solaris with multiple db w= riter, and finnaly set it
> to one, then the database run well.

I'm running with 3 DBWR= processes, for more then a year, with async I/O. Slowaris 9,
Oracle 9.2.0.5, RAC. My users are storing documents in= to the database as BLOB's and
that is very write-intensive, so I need multiple writers - 1 per I/O co= ntroller.
No problems. The only thing that is incompatible with async I/= O are I/O slaves, which
are intended to simulate async I/O. That is expl= ained in the note=20 97291.1.
However, on HP-UX  it is not recommended to use multi= ple DBWR processes and async I/O.
This is an old information, from the t= ime that I was working for OXHP, which used to
have 4-way OPS (not RAC) = system. I remember oracle support guy telling me that it is
not advised to use that combo on HP-UX 11.0. It was a brand new system,= we finished
transition from an old Siemens Reliant Unix to HP-UX on Sun= day, 09/09/2001. Tomorrow,
on Monday 09/10/2001 I was very busy solving = various problems, but everything was successful
in the end. I thought that I will relax a bit as of the day after. The = day after was, of
course, Tuesday 9/11/2001 and the world as I knew it e= nded that day.

--
Mladen Gogala
http://www.mgogala.com



------=_Part_1925_27403192.1126357652179-- -- http://www.freelists.org/webpage/oracle-l