I've had ZFS + NFS combo on a 4x 200MHz (yes, MegaHertz!) Pentium Prosin 
PRODUCTION since July 2006 (Solaris 10 u2 (6/06)), and I get 12MB/s ona striped 
RAID-Z (2 stripes x 3 disks per RAID-Z stripe).
I think 12MB is pretty damn good performance considering the system onlyhas 
768MB of RAM and 200MHz processors, so you and I will have to disagreeregarding 
your statement that "NFS is just unusable for everyday use".
That server has served out and written terabytes and terabytes of data in 
thelast four years.
And this is on ZFS on-disk revision 2, which is ancient by today's 
standards(ZFS on-disk format is now at revision 14 or higher).
But it helps to understand what is going on behind the scenes. I think 
thatarticle from Roch will help. You expect to have your cake and eat it too, 
whichis still a no-go, even in this day and age.
Date: Fri, 29 Jan 2010 09:44:15 +0100
From: [email protected]
To: ug-chosug at opensolaris.org
Subject: Re: [ug-chosug] [NFS] Zfs sharenfs bad performance

Hello Gerald,

Thanks for your reply. I also tried to sharenfs=off and sharing with share 
command. It gives me the same result.
For a bit mote details, the zpool is a raid10 array of 4x1TB hard drives. it's 
running snv_129 with dedup on - deactivating it doesn't change jack.


Zipping files is not what i'd call a workaround... NFS is just unusable for 
everyday use :(


Olivier

On Fri, Jan 29, 2010 at 9:22 AM,  <gerald.eggenberger at sunwave.ch> wrote:

Hi Olivier



I know that there are some NFS-Problems:

http://opensolaris.org/jive/thread.jspa?messageID=433708



The same Problem exists with Solaris 10/09.



It works, but performance is poor. If you copy a lot of little files it's even 
worse. Therefore i think the problem has to do with File-Attribute settings and 
not with network-performance. Copy less big files an the performance should be 
'ok'.




As workaround ZIP all files before copy and extract it on the target system.



Cheers G?rald













Zitat von Mauras Olivier <oliver.mauras at gmail.com>:




Hi there,



When i say bad, this is really bad... I'm sharing a dataset trough zfs

sharenfs. I connected a MacOSX client, and tested reads and writes.

Performances are far below what i'd expect from my gigabit network... Every

writes and deletion takes ages to complete...



To isolate the problem i fired up a script that creates 10000 100k files and

delete them. Running this locally gives 0.400s for deletion and 24s for

total time - creation + deletion.

I actually locally mounted the NFS to see if this could come from OSX... And

this is bad, the same script takes 8minutes over NFS :( - a bit worse from

OSX client



I also share another dataset from the same pool trough SMB and it's smoothly

running at ~48MB/s

I'm open to every idea. I already did a lot of inspection and nothing fancy

got reported with snoop or nfsstat.







Regards,



Olivier








_______________________________________________

Switzerland OpenSolaris Mailing List

ug-chosug at opensolaris.org

http://mail.opensolaris.org/mailman/listinfo/ug-chosug

wiki: http://wikis.sun.com/display/chosug

LinkedIn: http://www.linkedin.com/groups?gid=161188&trk=hb_side_g

ISO images: http://mirror.switch.ch/ftp/mirror/opensolaris.org/release_isos/

IPS: http://ips.osug.ch:10000/release/

Pictures: http://www.flickr.com/photos/31224908 at N08


                                          
_________________________________________________________________
Hotmail: Trusted email with powerful SPAM protection.
https://signup.live.com/signup.aspx?id=60969
-------------- next part --------------
An HTML attachment was scrubbed...
URL: 
<http://mail.opensolaris.org/pipermail/ug-chosug/attachments/20100129/585d766a/attachment.html>

Reply via email to