All of lore.kernel.org
 help / color / mirror / Atom feed
From: "M. Todd Smith" <todd@sohovfx.com>
To: nfs@lists.sourceforge.net
Subject: Re: NFS tuning - high performance throughput.
Date: Tue, 14 Jun 2005 16:38:25 -0400	[thread overview]
Message-ID: <42AF4041.8030504@sohovfx.com> (raw)
In-Reply-To: <482A3FA0050D21419C269D13989C611308539C89@lavender-fe.eng.netapp.com>

Charles,

I've tried mounting both TCP and UDP and with rwsizes varying from 8k to 
32k, 32k rwsizes are giving us the best performance thus far.

Indeed I did skip a level with the ping times didn't I, 200 microseconds 
it is.

Cheers
Todd


Lever, Charles wrote:

>first place i would look is mount options on the client.  are you
>mounting with NFS over TCP, or NFS over UDP?
>
>(btw, your ping time should be 200 microseconds, not 200 nanoseconds).
>
>  
>
>>-----Original Message-----
>>From: M. Todd Smith [mailto:todd@sohovfx.com] 
>>Sent: Tuesday, June 14, 2005 4:18 PM
>>To: nfs@lists.sourceforge.net
>>Subject: [NFS] NFS tuning - high performance throughput.
>>
>>
>>Hello all,
>>
>>I've been attempting to get a better hold over our NFS performance as 
>>our network grows and grows.
>>
>>We've recently upgraded our NFS server machine to a dual 3.2ghz Xeon 
>>running Fedora Core 3 (kernel 2.6.11-1.14_FC3smp) w/ 4Gb RAM. 
>> Coupled 
>>with this machine we have a 2 Broadcom NetExtreme 2 Port PCI-X NIC on 
>>its own PCI-X bus (133Mhz).  Attached to this machine is our fibre 
>>channel SAN, using Seagate fibre-channel drives and an LSI 
>>dual channel 
>>2gigabit fibre adapter on its own PCI-X bus.  Local RW is 
>>~135Mbytes/sec.
>> 
>>The 4 ports are trunked together using bonding (balanced round robin 
>>mode), and trunked together on our Extreme Networks Summit 
>>400i switch.  
>>All the test machines are attached to this switch, making everything 
>>within one hop and ping times of less than a 200 nanoseconds.
>>
>>Current test machine is running Suse 9.2 and has an Intel 100/1000 XT 
>>server adapter (e1000 driver) on a shared but not high 
>>traffic PCI bus.  
>>Other test machines include some Dell PowerEdge 1850's with onboard 
>>Intel NICs, and some Apple G4, G5 and Xserves.
>>
>>I have read most of the tuning guides I can find on the net and 
>>attempted just about everything I can get my hands on (I have 
>>not tried 
>>jumbo frames yet, still waiting for some downtime to attempt 
>>that).  My 
>>problem is that no matter how I tune the machines I can get at max 
>>45Mb/ps throughput on NFS.  This was the same throughput we 
>>were getting 
>>with our old server with PCI cards, moreover this throughput 
>>is roughly 
>>the same for every machine on our network.  Theoretically we 
>>should be 
>>able to get much higher values.
>>
>>Any idea as to why this is?  I can provide config files and such if 
>>needed, but I'm really at a loss as to where to start.
>>
>>Cheers
>>Todd
>>
>>-- 
>>Systems Administrator
>>----------------------------------
>>Soho VFX - Visual Effects Studio
>>99 Atlantic Avenue, Suite 303
>>Toronto, Ontario, M6K 3J8
>>(416) 516-7863 
>>http://www.sohovfx.com
>>----------------------------------
>>
>>
>> 
>>
>>
>>
>>-------------------------------------------------------
>>SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
>>from IBM. Find simple to follow Roadmaps, straightforward articles,
>>informative Webcasts and more! Get everything you need to get up to
>>speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
>>_______________________________________________
>>NFS maillist  -  NFS@lists.sourceforge.net
>>https://lists.sourceforge.net/lists/listinfo/nfs
>>
>>    
>>
>
>
>  
>


-- 
Systems Administrator
----------------------------------
Soho VFX - Visual Effects Studio
99 Atlantic Avenue, Suite 303
Toronto, Ontario, M6K 3J8
(416) 516-7863 
http://www.sohovfx.com
----------------------------------


 



-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

       reply	other threads:[~2005-06-14 20:38 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <482A3FA0050D21419C269D13989C611308539C89@lavender-fe.eng.netapp.com>
2005-06-14 20:38 ` M. Todd Smith [this message]
2005-06-15  1:56   ` NFS tuning - high performance throughput Dan Stromberg
2005-06-14 20:40 Lever, Charles
     [not found] <20050610031144.4B9CA12F8C@sc8-sf-spam2.sourceforge.net>
2005-06-14 20:17 ` M. Todd Smith
2005-06-14 20:41   ` Bill Rugolsky Jr.
2005-06-14 22:49     ` M. Todd Smith
2005-06-15 13:03       ` Roger Heflin
2005-06-15 14:47         ` M. Todd Smith
2005-06-15 15:28           ` Roger Heflin
2005-06-15 19:13             ` Dan Stromberg
2005-06-15 19:52               ` Roger Heflin
2005-06-15 20:11                 ` Dan Stromberg
2005-06-15 20:31                   ` Roger Heflin
2005-06-15 20:33                   ` Chris Penney
2005-06-15 17:47       ` Bill Rugolsky Jr.
2005-06-15 20:33         ` M. Todd Smith
2005-06-15 22:43           ` Bill Rugolsky Jr.
2005-06-15 22:47         ` Greg Banks
2005-06-14 20:50   ` Bill Rugolsky Jr.
2005-06-14 21:04   ` Chris Penney
2005-06-14 21:06     ` Chris Penney
2005-06-14 21:11   ` Roger Heflin

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=42AF4041.8030504@sohovfx.com \
    --to=todd@sohovfx.com \
    --cc=nfs@lists.sourceforge.net \
    /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.