linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Max Waterman <davidmaxwaterman+kernel@fastmail.co.uk>
To: Phillip Susi <psusi@cfl.rr.com>
Cc: linux-kernel@vger.kernel.org
Subject: Re: io performance...
Date: Fri, 20 Jan 2006 13:58:32 +0800	[thread overview]
Message-ID: <43D07C08.5000903@fastmail.co.uk> (raw)
In-Reply-To: <43CE5C7A.5060608@cfl.rr.com>

Phillip Susi wrote:
> Right, the kernel does not know how many disks are in the array, so it 
> can't automatically increase the readahead.  I'd say increasing the 
> readahead manually should solve your throughput issues.

Any guesses for a good number?

We're in RAID10 (2+2) at the moment on 2.6.8-smp. These are the block 
numbers I'm getting using bonnie++ :

ra	wr	rd
256	68K	46K
512	67K	59K
640	67K	64K
1024	66K	73K
2048	67K	88K
3072	67K	91K
8192	71K	96K
9216	67K	92K
16384	67K	90K

I think we might end up going for 8192.

We're still wondering why rd performance is so low - seems to be the 
same as a single drive. RAID10 should be the same performance as RAID0 
over two drives, shouldn't it?

Max.

> 
> Max Waterman wrote:
>>
>> I left the stripe size at the default, which, I believe, is 64K bytes; 
>> same as your fakeraid below.
>>
>> I did play with 'blockdev --setra' too.
>>
>> I noticed it was 256 with a single disk, and, with s/w raid, it 
>> increased by 256 for each extra disk in the array. IE for the raid 0 
>> array with 4 drives, it was 1024.
>>
>> With h/w raid, however, it did not increase when I added disks. Should 
>> I use 'blockdev --setra 320' (ie 64 x 5 = 320, since we're now running 
>> RAID5 on 5 drives)?
>>
> 


  reply	other threads:[~2006-01-20  5:58 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-01-16  7:35 io performance Max Waterman
2006-01-16  7:32 ` Jeff V. Merkey
2006-01-17 13:57   ` Jens Axboe
2006-01-17 19:17     ` Jeff V. Merkey
2006-01-16  8:35 ` Pekka Enberg
2006-01-17 17:06 ` Phillip Susi
2006-01-18  7:24   ` Max Waterman
2006-01-18 15:19     ` Phillip Susi
2006-01-20  5:58       ` Max Waterman [this message]
2006-01-20 13:42         ` Ian Soboroff
2006-01-25  6:36           ` Max Waterman
2006-01-25 14:19             ` Ian Soboroff
2006-01-25 13:09           ` Bernd Eckenfels
2006-01-18  3:02 ` Max Waterman
2006-01-18  4:30   ` Jeff V. Merkey
2006-01-18  5:09     ` Max Waterman
2006-01-18  4:37       ` Jeff V. Merkey
2006-01-18  7:06         ` Max Waterman
2006-01-18  9:21     ` Alan Cox
2006-01-18 15:48       ` Phillip Susi
2006-01-18 16:25         ` Bartlomiej Zolnierkiewicz
2006-01-19  0:48 ` Adrian Bunk
2006-01-19 13:18   ` Max Waterman
     [not found] <5vx8f-1Al-21@gated-at.bofh.it>
     [not found] ` <5wbRY-3cF-3@gated-at.bofh.it>
     [not found]   ` <5wdKh-5wF-15@gated-at.bofh.it>
2006-01-19  1:58     ` Robert Hancock
2006-01-19 13:14       ` Max Waterman
2006-01-19 14:08         ` Alan Cox
2006-01-20  4:09           ` Max Waterman
2006-01-20  4:27             ` Alexander Samad
2006-01-20 12:52             ` Alan Cox
2006-01-19 11:39 Al Boldi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=43D07C08.5000903@fastmail.co.uk \
    --to=davidmaxwaterman+kernel@fastmail.co.uk \
    --cc=linux-kernel@vger.kernel.org \
    --cc=psusi@cfl.rr.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).