Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: Raw Devices v.s. File System

RE: Raw Devices v.s. File System

From: LoughMiller, Gregory <Gregory_LoughMiller_at_bscc.bls.com>
Date: Wed, 12 Jul 2000 07:18:01 -0400
Message-Id: <10556.111780@fatcity.com>


We have built, designed, managed and supported medium to large scale(200GB upto 1.1 TB) databases for 5 years.

Every time we start a project-the question arises "raw or cooked"? We have tried raw, cooked, and even the Veritas product of Quick I/O-all on EMC disks... In some cases we found that RAW was quicker than cooked with the Veritas Quick I/O. In some cases it wasn't. But both RAW and QUICK I/O configurations had a better performance than just the straight cooked file systems.

It becomes a personal opinion and what you are comfortable with in managing the environment. We have a very good working relationship with the Sys admins-and we are both comfortable working with raw devices.. So we typically go with RAW devices.. We also used the Verita Volume Manager, which made the management of the raw devices easier to manage...

So the things to consider:

1. How comfortable are you with raw devices
2. What type of relationship do you have with the sys admins
3. Build a strategy for the use of either cooked or raw devices...
4. Take the time to build a "lab"-test each scenario with your application
and see what works best..
5. And the disk vendor that the shop uses(EMC, SUN, MTI,etc...)

greg

-----Original Message-----
From: Gaja Krishna Vaidyanatha [mailto:gajav_at_yahoo.com] Sent: Tuesday, July 11, 2000 5:22 PM
To: Multiple recipients of list ORACLE-L Subject: Re: Raw Devices v.s. File System

Michael,

If you are not running Oracle Parallel Server, believe me when I tell you that you will get comparable performance using Veritas vxfs file systems as compared to Raw Devices. There have been enough benchmarks and real-life production configurations that have been done to prove this. Been there...done that. It is not worth dealing with "raw devices" if you have an advanced file system such as Veritas with an I/O driver that facilitates "non-buffered I/O". This takes care of the "double buffering" problem that is prevelant with normal ufs filesystems.

In addition to configuring vxfs filesystems you need to install and configure the "Database Accelerator" - "Quick I/O for Oracle" component of Veritas. This simulates the effects of "non-buffered I/O". Basically, while configuring the file systems, the Quick I-O tag is put on it (so to speak) and this assists the driver in identifying "Quick I/O" calls to that filesystem and intercepting it. It does it by "bypassing" the file system buffercache and performing it directly from the device. So you have "raw device" comparable performance, without sacrificing the benefits of a cooked filesystem.

To minimize the overhead in filesystem-database block size mapping, keep the vxfs filesystem block size (default 8k) and the db_block_size of the database to the same. Then configure your Oracle database, just as usual. Leave the rest to Quick I/O, it does pretty well. There are white papers at www.veritas.com on Quick I/O and Oracle and how it works.

Cheers,

Gaja.


Gaja Krishna Vaidyanatha
Director, I-O Management Products
Quest Software Inc.
(972)-304-1170
gajav_at_yahoo.com

"Opinions and views expressed are my own and not of Quest"



Do You Yahoo!?
Get Yahoo! Mail - Free email you can access from anywhere! http://mail.yahoo.com/
-- 
Author: Gaja Krishna Vaidyanatha
  INET: gajav_at_yahoo.com

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
Received on Wed Jul 12 2000 - 06:18:01 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US