My understanding of "Cooked device" is right ?
Greeting !! when I like to use cooked device for informix chunk , usually I touch a file in filesystem , said /home/informix/chunk1.dat and it is used as informix chunk ... The 2'nd way to use cooked device for informix chunk is to run fdisk to add a new partition , said /dev/sda10 and then use it directly to be informix's chunk !! I never use the 2'nd way , I wonder if I should do mkfs -t ext3 /dev/sda10 first , and then use sda10 as informix chunk , or I should not do mkfs at all .
If you are using the block device as a cooked chunk you need not run mkfs because Informix will overwrite all of the filesystem's infrastructure with its own structures when it creates the reserved pages at the beginning of the chunk. Art Art S. Kagel Advanced DataTools (www.advancedatatools.com) Blog: http://informix-myview.blogspot.com/ Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on my employer, Advanced DataTools, the IIUG, nor any other organization with which I am associated either explicitly, implicitly, or by inference. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Wed, Nov 23, 2011 at 10:50 PM, MARS CHEN <hedgezzz@yahoo.com.tw> wrote: > Greeting !! > > when I like to use cooked device for informix chunk , > usually I touch a file in filesystem , said /home/informix/chunk1.dat > and it is used as informix chunk ... > > The 2'nd way to use cooked device for informix chunk > is to run fdisk to add a new partition , said /dev/sda10 > and then use it directly to be informix's chunk !! > > I never use the 2'nd way , I wonder if I should do > mkfs -t ext3 /dev/sda10 first , and then use sda10 as informix chunk , > or I should not do mkfs at all . > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --20cf303ddac2177fc904b273beb2
Thank you , Art !! I've read several webpages that using a ext3 , ext4 file as I mentioned /home/informix/chunk.dat is not a good idea , that means , if I can not use raw device as chunk for some reasons , use cooked device like my examply /dev/sda10 is a better idea than use ext3 or ext4 file as informix's chunk ... that is because /dev/sda10 is not formated as ext3 or ext4 , so it has no journaling capability , journaling capability harm IO performance of informix , am I correct ?
Correct. I'm actually doing some testing of various chunk configurations with IBM because of some performance problems we discovered under Linux. There seems to be a problem in the IO subsystems in the latest Linux kernels with write performance being from 3x to 4x faster than reading. So far COOKED (ie block) device is fastest for reading and RAW (character) device is fastest for writing but only by about 4%. Direct_IO doesn't seem to make much of a difference on Linux with the latest kernels. For writing, RAW is about 55% faster than EXT2 and more than 2x as fast as EXT3 or EXT4 (slowest). Because of the kernel performance problems I don't have good figures for reading, but we are testing older kernels as well and on them the read performance is very similar to the write performance and similar to write performance on the newer kernel. These tests were conducted with a simple C test program, not with the Informix engine, so we are testing the OS and various chunk configurations. Also testing was done with a single PC class disk drive with less cache than the data being read or written (~320MB non-sequential). Art Art S. Kagel Advanced DataTools (www.advancedatatools.com) Blog: http://informix-myview.blogspot.com/ Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on my employer, Advanced DataTools, the IIUG, nor any other organization with which I am associated either explicitly, implicitly, or by inference. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Thu, Nov 24, 2011 at 12:05 AM, MARS CHEN <hedgezzz@yahoo.com.tw> wrote: > Thank you , Art !! > > I've read several webpages that using a ext3 , ext4 file as > I mentioned /home/informix/chunk.dat is not a good idea , > that means , if I can not use raw device as chunk for some reasons , > use cooked device like my examply /dev/sda10 is a better idea > than use ext3 or ext4 file as informix's chunk ... > > that is because /dev/sda10 is not formated as ext3 or ext4 , > so it has no journaling capability , journaling capability harm IO > performance of informix , am I correct ? > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --14dae93411f96c47a404b2746243
Hi Art. When you say 'latest kernels' - can you be specific as to the version number - and vendor if that's relevant? We're pretty heavy linux users over here, and I'd like to be able to either confirm wether this affects any of our environments. Which IO scheduler where you using? For reference, when I last tested, my benchmarks had RAW devices as about 10-20% faster than cooked files on linux for writes for Informix, although I was testing with ISCSI and FC Raid-10 devices at the time, the smallest of which would have been 6 disks - and I suspect that makes quite a bit of difference for journaled workloads in particular. Raw was a little quicker for reads to - there appeared to be less seek activity on the drives. Thanks, Tristan Tristan Ball - Hosted Services Manager VIC Pronto Hosted Services 20 Lakeside Drive, Burwood East, VIC 3151 Phone: +61 3 9887 7770 | Email: tristanb@pronto.com.au Mobile: +61 408 397 473 For PHS helpdesk support, please email phs@pronto.com.au For urgent after hours support phone: 1800 622 556 ---Legal Notice--- The email message and any attachments are confidential and subject to copyright. If you are not the intended recipient, any use, interference with, disclosure or copying of this material is unauthorised and prohibited. No part may be reproduced, adapted or transmitted without the written permission of the copyright owner. If you have received this email in error, please immediately advise the sender by return email and delete the message from your system. Before opening or using attachments, check for viruses and defects. Our liability is limited to re-supplying any affected attachments. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Art Kagel Sent: Thursday, 24 November 2011 4:34 PM To: ids@iiug.org Subject: Re: My understanding of "Cooked device" is right ? [25469] Correct. I'm actually doing some testing of various chunk configurations with IBM because of some performance problems we discovered under Linux. There seems to be a problem in the IO subsystems in the latest Linux kernels with write performance being from 3x to 4x faster than reading. So far COOKED (ie block) device is fastest for reading and RAW (character) device is fastest for writing but only by about 4%. Direct_IO doesn't seem to make much of a difference on Linux with the latest kernels. For writing, RAW is about 55% faster than EXT2 and more than 2x as fast as EXT3 or EXT4 (slowest). Because of the kernel performance problems I don't have good figures for reading, but we are testing older kernels as well and on them the read performance is very similar to the write performance and similar to write performance on the newer kernel. These tests were conducted with a simple C test program, not with the Informix engine, so we are testing the OS and various chunk configurations. Also testing was done with a single PC class disk drive with less cache than the data being read or written (~320MB non-sequential). Art Art S. Kagel Advanced DataTools (www.advancedatatools.com) Blog: http://informix-myview.blogspot.com/ Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on my employer, Advanced DataTools, the IIUG, nor any other organization with which I am associated either explicitly, implicitly, or by inference. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Thu, Nov 24, 2011 at 12:05 AM, MARS CHEN <hedgezzz@yahoo.com.tw> wrote: > Thank you , Art !! > > I've read several webpages that using a ext3 , ext4 file as I > mentioned /home/informix/chunk.dat is not a good idea , that means , > if I can not use raw device as chunk for some reasons , use cooked > device like my examply /dev/sda10 is a better idea than use ext3 or > ext4 file as informix's chunk ... > > that is because /dev/sda10 is not formated as ext3 or ext4 , so it has > no journaling capability , journaling capability harm IO performance > of informix , am I correct ? > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --14dae93411f96c47a404b2746243 ******************************************************************************* Forum Note: Use "Reply" to post a response in the discussion forum.