All of lore.kernel.org
 help / color / mirror / Atom feed
From: Tommy Apel Hansen <tommyapeldk@gmail.com>
To: linux-raid@vger.kernel.org
Subject: RAID5 performace issue
Date: Thu, 22 Nov 2012 21:53:54 +0100	[thread overview]
Message-ID: <1767331.K4nRMCSJWv@workstation-home> (raw)

Hello list,
I'm experiencing some issus with RAID5 and 5 or more disks, while doign 
sequencial 4k writes to and array with 5 or more disks I'm only getting 
~279IOPs where when using 4 disks yields ~2752IOPs, it's a 1/10 in IOPs but 
with more disks so the result should actually be better the more disks there 
is, see results and test below.

Regards
/Tommy

RAID5 5 disks:
echo 16384 > /sys/block/md0/md/stripe_cache_size
mkfs.xfs -l su=32k -d su=512k,sw=4 -f /dev/md0
mdadm -D /dev/md0:
        Version : 1.2
  Creation Time : Wed Nov 21 20:36:46 2012
     Raid Level : raid5
     Array Size : 7814051840 (7452.06 GiB 8001.59 GB)
  Used Dev Size : 1953512960 (1863.02 GiB 2000.40 GB)
   Raid Devices : 5
  Total Devices : 5
    Persistence : Superblock is persistent

    Update Time : Thu Nov 22 20:59:27 2012
          State : clean 
 Active Devices : 5
Working Devices : 5
 Failed Devices : 0
  Spare Devices : 0

         Layout : right-asymmetric
     Chunk Size : 512K
fio sequential write ioengine=sync, iodepth=1 :
-- 512b bw=2207.3KB/s, iops=4414
-- 1k      bw=3586.4KB/s, iops=3586
-- 2k      bw=5584.7KB/s, iops=2792
-- 4k      bw=1119.6KB/s, iops=279
-- 8k      bw=2107.4KB/s, iops=263
-- 16k    bw=4438.3KB/s, iops=277
-- 32k    bw=6826.7KB/s, iops=213
-- 64k    bw=9167.5KB/s, iops=143
-- 128k  bw=16580KB/s, iops=129

fio sequential read ioengine=sync, iodepth=1:
-- 512b bw=14069KB/s, iops=28137
-- 1k      bw=20511KB/s, iops=20511
-- 2k      bw=25173KB/s, iops=12586
-- 4k      bw=30654KB/s, iops=7663
-- 8k      bw=47503KB/s, iops=5937
-- 16k    bw=59487KB/s, iops=3717
-- 32k    bw=72496KB/s, iops=2265
-- 64k    bw=80117KB/s, iops=1251
-- 128k  bw=88286KB/s, iops=689

RAID5 4 disks:
mkfs.xfs -l su=32k -d su=512k,sw=3 -f /dev/md0
mdadm -D /dev/md0
        Version : 1.2
  Creation Time : Thu Nov 22 21:29:45 2012
     Raid Level : raid5
     Array Size : 5860538880 (5589.05 GiB 6001.19 GB)
  Used Dev Size : 1953512960 (1863.02 GiB 2000.40 GB)
   Raid Devices : 4
  Total Devices : 4
    Persistence : Superblock is persistent

    Update Time : Thu Nov 22 21:35:28 2012
          State : clean 
 Active Devices : 4
Working Devices : 4
 Failed Devices : 0
  Spare Devices : 0

         Layout : right-asymmetric
     Chunk Size : 512K
fio sequential write ioengine=sync, iodepth=1 :
-- 512b bw=2553.2KB/s, iops=5107
-- 1k      bw=5486.9KB/s, iops=5486
-- 2k      bw=10785KB/s, iops=5392
-- 4k      bw=11012KB/s, iops=2752
-- 8k      bw=15316KB/s, iops=1914
-- 16k    bw=15933KB/s, iops=995
-- 32k    bw=17368KB/s, iops=542
-- 64k    bw=20633KB/s, iops=322
-- 128k  bw=26263KB/s, iops=205

fio sequential read ioengine=sync, iodepth=1:
-- 512b bw=14112KB/s, iops=28224
-- 1k      bw=20512KB/s, iops=20512
-- 2k      bw=25199KB/s, iops=12599
-- 4k      bw=30438KB/s, iops=7609
-- 8k      bw=48065KB/s, iops=6008
-- 16k    bw=68692KB/s, iops=4293
-- 32k    bw=82708KB/s, iops=2584
-- 64k    bw=96403KB/s, iops=1506
-- 128k bw=111824KB/s, iops=873


                 reply	other threads:[~2012-11-22 20:53 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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=1767331.K4nRMCSJWv@workstation-home \
    --to=tommyapeldk@gmail.com \
    --cc=linux-raid@vger.kernel.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.