mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Nagy Tibor <nagyt@otpbank.hu>
To: linux-kernel@vger.kernel.org
Subject: Dell Perc 3/QC performance
Date: Wed, 21 Aug 2002 16:26:04 +0200	[thread overview]
Message-ID: <3D63A2FC.92BA1342@otpbank.hu> (raw)

[-- Attachment #1: Type: text/plain, Size: 3444 bytes --]

Hi,

I am not satisfied with the disk I/O performance of the RAID controller
'Dell Perc 3/QC' (AMI megaraid) under Linux.

Short description of my problem:

We have a Dell PowerEdge 6600 + PERC 3/QC, and I am not satisfied with
the performance. I/O speed is not linearly growing if I read more disks
simultenausly. If I read 1 disk, I get about 49 MB/s read speed, if I
read 2 disks simultaneosly I get about 64 MB/s read speed (only
+15MB/s), if I read 3 disks simultaneosly I get about 68 MB/s read speed
(only +4MB/s), If I read more disk I don't get higher speed.

Do you have any idea, what is the problem?

Detailed description:

1. Configuration

Dell Poweredge 6600: 4 x 1400 MHz, 4 GB RAM
PERC 3/QC Firmware: 1.61N Bios: 3.17

 Logical drive 0 (18 GB) - RAID 1 - Channel 0 Id 0 - 18 GB
                                  \ Channel 0 Id 1 - 18 GB

 Logical drive 1 (22 GB) - RAID 1 - Channel 2 Id 0 - 72 GB
 Logical drive 2 (22 GB) +        \ Channel 3 Id 9 - 72 GB
 Logical drive 3 (24 GB) /

 Logical drive 4 (22 GB) - RAID 1 - Channel 2 Id 1 - 72 GB
 Logical drive 5 (22 GB) +        \ Channel 3 Id 10 - 72 GB
 Logical drive 6 (24 GB) /

 Logical drive 7 (22 GB) - RAID 1 - Channel 2 Id 2 - 72 GB
 Logical drive 8 (22 GB) +        \ Channel 3 Id 11 - 72 GB
 Logical drive 9 (24 GB) /

 18 GB disks: Quantum ATLAS 10KIII 18 SCA (160 MB SCSI)
 72 GB disks: Hitachi DK32DJ-72 (160 MB SCSI)

 The 18 GB disks are installed in the server itself, the 72 GB disks are
installed in a PowerVault 221S disk tower with splitted backplane.

 The 72 GB disks are splitted into 22-24 GB logical drives, because
Informix database engine can use max 2GB chunks (partitions), and only
15 partitions can be installed in linux on 1 drive.

 'Cache policy' is set to 'Direct IO' and 'Read (ahead) policy' to
'adaptive' for all logical drives. (I tried other confugurations, too.)

 Logical drives 1-9 are used as raw devices by informix, logical drive 0
is for the operating system.

2. Operating system

SuSe Linux 8.0, kernel 2.4.18

emiir:~ # uname -a
Linux emiir 2.4.18-64GB-SMP #1 SMP Wed Mar 27 13:58:12 UTC 2002 i686
unknown


3. I/O test

3.1 Read from Logical drive 1

dd if=/dev/rsdb1 of=/dev/null bs=1024k	->  49 MB/s I/O

3.2 Read from Logical drive 1 and 4 (different disks)

dd if=/dev/rsdb1 of=/dev/null bs=1024k &
dd if=/dev/rsde1 of=/dev/null bs=1024k &  -> 64 MB/s I/O

3.2 Read from Logical drive 1, 4 and 7 (different disks)

dd if=/dev/rsdb1 of=/dev/null bs=1024k &
dd if=/dev/rsde1 of=/dev/null bs=1024k &
dd if=/dev/rsdh1 of=/dev/null bs=1024k &  -> 68 MB/s I/O

All I/O speed came from the column 'bi' of 'vmstat 1'.


4. Result

I do not understand, why the speed is not increased nearly linearly. The
I/O between the PERC controller and the host is 64 bit PCI with 66MHz,
ca. 530MB/s, each of the four SCSI channel has a speed of 160MB/s. At
least test 3.2 should result the double of the speed of test 3.1 (two
SCSI channel, separate disks).

I see also on the dell site 300 MB/s for sequential read:
  http://www.dell.com/us/en/biz/topics/power_ps1q02-perc.htm

Do you have any idea what can be wrong?

Than you

Tibor

------------------------------------------------------------------------
Tibor Nagy
National Savings and Commercial Bank Ltd (OTP Bank)
H-1051 Budapest Nador u. 16.
Tel: 00 36 1 374 6990	Fax: 00 36 1 374 6981	E-mail: nagyt@otpbank.hu
------------------------------------------------------------------------

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 2027 bytes --]

             reply	other threads:[~2002-08-21 14:13 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-08-21 14:26 Nagy Tibor [this message]
2002-08-21 15:09 Downey, Brian

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=3D63A2FC.92BA1342@otpbank.hu \
    --to=nagyt@otpbank.hu \
    --cc=linux-kernel@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®