Return-Path: Delivered-To: 2-oracle-l@orafaq.com Received: (qmail 20259 invoked from network); 6 Dec 2007 09:40:49 -0600 Received: from freelists-180.iquest.net (HELO turing.freelists.org) (206.53.239.180) by 69.64.49.119 with SMTP; 6 Dec 2007 09:40:48 -0600 Received: from localhost (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 1443C7D4489; Thu, 6 Dec 2007 10:40:47 -0500 (EST) Received: from turing.freelists.org ([127.0.0.1]) by localhost (turing.freelists.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 21776-03; Thu, 6 Dec 2007 10:40:46 -0500 (EST) Received: from turing (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 75FF17D4425; Thu, 6 Dec 2007 10:40:46 -0500 (EST) Received: with ECARTIS (v1.0.0; list oracle-l); Thu, 06 Dec 2007 09:53:50 -0500 (EST) Received: from localhost (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 4B8837D38F2 for ; Thu, 6 Dec 2007 09:53:50 -0500 (EST) Received: from turing.freelists.org ([127.0.0.1]) by localhost (turing.freelists.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 11083-08 for ; Thu, 6 Dec 2007 09:53:50 -0500 (EST) Received: from web35404.mail.mud.yahoo.com (web35404.mail.mud.yahoo.com [66.163.179.113]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with SMTP id C0A667D37ED for ; Thu, 6 Dec 2007 09:53:47 -0500 (EST) Received: (qmail 69658 invoked by uid 60001); 6 Dec 2007 14:53:46 -0000 X-YMail-OSG: dsxl1sQVM1kNqn.HOKS7FmlccBxURNCICpTxmtnY9ii376fxfe4cZrulj_xjhpG9SgaUUegy6CWDxIRZ0Z3hzY4owP3zKVweQ90K_CgAak6XS7w_gNk- Received: from [67.90.115.78] by web35404.mail.mud.yahoo.com via HTTP; Thu, 06 Dec 2007 06:53:45 PST Date: Thu, 6 Dec 2007 06:53:45 -0800 (PST) From: Dan Norris Subject: Re: Cluster File System Versus ASM for RAC Deployment in Production?... Pros & Cons To: Matthew Zito , VIVEK_SHARMA@infosys.com, oracle-l@freelists.org MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="0-1481225797-1196952825=:69643" Message-ID: <124157.69643.qm@web35404.mail.mud.yahoo.com> X-archive-position: 3748 X-ecartis-version: Ecartis v1.0.0 Sender: oracle-l-bounce@freelists.org Errors-to: oracle-l-bounce@freelists.org X-original-sender: dannorris@dannorris.com Precedence: normal Reply-to: dannorris@dannorris.com List-help: List-unsubscribe: List-software: Ecartis version 1.0.0 List-Id: oracle-l X-List-ID: oracle-l List-subscribe: List-owner: List-post: List-archive: X-list: oracle-l X-Virus-Scanned: Debian amavisd-new at localhost.localdomain --0-1481225797-1196952825=:69643 Content-Type: text/plain; charset=us-ascii Sorry--I forgot to also include that I completely agree with Matt's comments about using ASM + a CFS--especially to support a secondary archivelog destination. I sometimes also use a local (non-shared) filesystem for a secondary archivelog destination if a CFS isn't available. Some customers don't like to add the additional "special" software required to support the CFS in an already-complex environment. Good comments, Matt. Dan ----- Original Message ---- From: Dan Norris To: Matthew Zito ; VIVEK_SHARMA@infosys.com; oracle-l@freelists.org Sent: Thursday, December 6, 2007 8:34:46 AM Subject: Re: Cluster File System Versus ASM for RAC Deployment in Production?... Pros & Cons >> However, OCFS2 has no capability for growing filesystems, creating situations where using OCFS2 for datafiles means increasing the number of mounts over time, which gets messy. That's incorrect. "tunefs.ocfs2 -S /dev/sde1" will grow the FS to the size of the partition where it resides. Just tested it this week: [root@ch-srlxdb01 ~]# tunefs.ocfs2 -S /dev/sde1 tunefs.ocfs2 1.2.7 Changing volume size from 1309649 blocks to 1834541 blocks Proceed (y/N): y Resized volume Wrote Superblock ----- Original Message ---- From: Matthew Zito To: dannorris@dannorris.com; VIVEK_SHARMA@infosys.com; oracle-l@freelists.org Sent: Thursday, December 6, 2007 8:29:24 AM Subject: RE: Cluster File System Versus ASM for RAC Deployment in Production?... Pros & Cons I always recommend using both ASM and a CFS, where appropriate. For example, on Linux systems, where there is a free, Oracle-supported CFS (OCFS2), I recommend making a small OCFS2 filesystem (or two) for storage of OCR and voting files. Then, make the primary datafile/index/log storage on ASM. Optionally, they can then have a large dump or backup filesystem that is OCFS2, or NFS - but I always recommend keeping an archive location outside of ASM, so if the ASM instance won't start up, you can at least get to your archive logs, and presumably you're doing backups or a standby database somewhere. The reason for using both ASM and CFS on Linux is because dealing with multiple block devices for the various OCR and Voting devices is annoying and complex, and typically you'll end up wasting a lot of disk space (i.e. allocating an 8GB lun for a 100MB ocr device). By using a clustered file system, you can put multiple objects on the one disk, and if necessary, store other things there. However, OCFS2 has no capability for growing filesystems, creating situations where using OCFS2 for datafiles means increasing the number of mounts over time, which gets messy. ASM solves that problem for you by doing very basic striping. Thanks, Matt -- Matthew Zito Chief Scientist GridApp Systems P: 646-452-4090 mzito@gridapp.com http://www.gridapp.com --0-1481225797-1196952825=:69643 Content-Type: text/html; charset=us-ascii
Sorry--I forgot to also include that I completely agree with Matt's comments about using ASM + a CFS--especially to support a secondary archivelog destination. I sometimes also use a local (non-shared) filesystem for a secondary archivelog destination if a CFS isn't available. Some customers don't like to add the additional "special" software required to support the CFS in an already-complex environment.

Good comments, Matt.

Dan

----- Original Message ----
From: Dan Norris <dannorris@dannorris.com>
To: Matthew Zito <mzito@gridapp.com>; VIVEK_SHARMA@infosys.com; oracle-l@freelists.org
Sent: Thursday, December 6, 2007 8:34:46 AM
Subject: Re: Cluster File System Versus ASM for RAC Deployment in Production?... Pros & Cons

>> However, OCFS2 has no capability for growing filesystems, creating situations where using OCFS2 for datafiles means increasing the number of mounts over time, which gets messy.

That's incorrect. "tunefs.ocfs2 -S /dev/sde1" will grow the FS to the size of the partition where it resides. Just tested it this week:

[root@ch-srlxdb01 ~]# tunefs.ocfs2 -S /dev/sde1

tunefs.ocfs2 1.2.7

Changing volume size from 1309649 blocks to 1834541 blocks

Proceed (y/N): y

Resized volume

Wrote Superblock



----- Original Message ----
From: Matthew Zito <mzito@gridapp.com>
To: dannorris@dannorris.com; VIVEK_SHARMA@infosys.com; oracle-l@freelists.org
Sent: Thursday, December 6, 2007 8:29:24 AM
Subject: RE: Cluster File System Versus ASM for RAC Deployment in Production?... Pros & Cons


I always recommend using both ASM and a CFS, where appropriate.  For example, on Linux systems, where there is a free, Oracle-supported CFS (OCFS2), I recommend making a small OCFS2 filesystem (or two) for storage of OCR and voting files.  Then, make the primary datafile/index/log storage on ASM.  Optionally, they can then have a large dump or backup filesystem that is OCFS2, or NFS - but I always recommend keeping an archive location outside of ASM, so if the ASM instance won't start up, you can at least get to your archive logs, and presumably you're doing backups or a standby database somewhere. 

The reason for using both ASM and CFS on Linux is because dealing with multiple block devices for the various OCR and Voting devices is annoying and complex, and typically you'll end up wasting a lot of disk space (i.e. allocating an 8GB lun for a 100MB ocr device).  By using a clustered file system, you can put multiple objects on the one disk, and if necessary, store other things there.  However, OCFS2 has no capability for growing filesystems, creating situations where using OCFS2 for datafiles means increasing the number of mounts over time, which gets messy.  ASM solves that problem for you by doing very basic striping.

Thanks,
Matt

--
Matthew Zito
Chief Scientist
GridApp Systems
P: 646-452-4090
mzito@gridapp.com
http://www.gridapp.com




--0-1481225797-1196952825=:69643-- -- http://www.freelists.org/webpage/oracle-l