🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

🏠 Home 📰 News 🗂️ Usenet Search 🔑 Login 📝 Register
Thread View: comp.os.linux.networking
8 messages
[T] Back to Threads [V] Tree View [G] Group Articles
8 total messages Started by WEILL Philippe Thu, 27 Jun 1996 00:00
NFS Performance between Linux and OSF1
#734
Author: WEILL Philippe
Date: Thu, 27 Jun 1996 00:00
43 lines
2660 bytes 
When I use my OSF1 box as a server for my linux (i586) box 
,performance is pretty bad

OSF1 sirocco V3.2 41 alpha
Dec 3000/600 server with 64M memory

Linux blizzard 2.0.0 #7 Wed Jun 12 16:57:00 MET DST 1996 i586
(NFSserver 2.2beta16)
pentium 90 overclock to 100 16Mo 1Go disk eide quantum fireball ,Motherboard zap
pa
3C590 Network card with 0.25 driver
Slacware 3.0

all test are made with the sames partions on both box with 20Mo file

result are also the same with NFSserver2.2beta4 on linux box 

Linux Client READ from OSF1 Server : ~350Ko/s
Linux Client Write to  OSF1 Server : ~30Ko/s
                                     ^^^^^^^
ftp linux to osf1 the same file    : ~900Ko/s 
                                     ^^^^^^^
OSF1 Client Read from Linux server : ~800Ko/s
OSF1 Client write  to Linux Server : ~500Ko/s

HP-UX9 Client Read from OSF1 server: ~500Ko/s
HP-UX9 Client Write to OSF1 server : ~400Ko/s

Linux Client READ from HP-UX9 Server : ~600Ko/s
Linux Client Write to  HP-UX9 Server : ~500Ko/s

Have you any advice to improve NFS write performance from linux to OSF1

-- 
###### L'informatique est monde logique qui ignore la logique humaine ######
|--------------------------------------------------------------------------|
| Weill Philippe                       |  CNRS Service Aeronomie           |
|                                      |  Universite Pierre et Marie Curie |
| Tel: +33-1-44274759                  |  Tour 15 Couloir 15-14            |
| Fax: +33-1-44273776                  |  4 Place Jussieu                  |
|                                      |  75252 Paris Cedex 05             |
| Email:philippe.weill@aero.jussieu.fr |  FRANCE                           |
|--------------------------------------------------------------------------|
Re: NFS Performance between Linux and OSF1
#735
Author: WEILL Philippe
Date: Thu, 27 Jun 1996 00:00
24 lines
2152 bytes 
I wrote:
> Linux Client READ from OSF1 Server : ~350Ko/s
> Linux Client Write to  OSF1 Server : ~30Ko/s
>                                      ^^^^^^^
> ftp linux to osf1 the same file    : ~900Ko/s
>
> Have you any advice to improve NFS write performance from linux to OSF1
>
using mount option size=8192 wsize=8192 is better
Linux Client READ from OSF1 Server : ~400Ko/s
Linux Client Write to  OSF1 Server : ~150Ko/s
                                     ^^^^^^^

Do you know  if it's possible to have better results or if it's the maximun
with the actual version of NFS Client for linux? 

|--------------------------------------------------------------------------|
| Weill Philippe                       |  CNRS Service Aeronomie           |
|                                      |  Universite Pierre et Marie Curie |
| Tel: +33-1-44274759                  |  Tour 15 Couloir 15-14            |
| Fax: +33-1-44273776                  |  4 Place Jussieu                  |
|                                      |  75252 Paris Cedex 05             |
| Email:philippe.weill@aero.jussieu.fr |  FRANCE                           |
|--------------------------------------------------------------------------|
Re: NFS Performance between Linux and OSF1
#733
Author: Juergen Holm
Date: Mon, 01 Jul 1996 00:00
50 lines
3088 bytes 
WEILL Philippe (philippe.weill@aero.jussieu.fr) wrote:
> I wrote:
> > Linux Client READ from OSF1 Server : ~350Ko/s
> > Linux Client Write to  OSF1 Server : ~30Ko/s
> >                                      ^^^^^^^
> > ftp linux to osf1 the same file    : ~900Ko/s
> >
> > Have you any advice to improve NFS write performance from linux to OSF1
> >
> using mount option size=8192 wsize=8192 is better
> Linux Client READ from OSF1 Server : ~400Ko/s
> Linux Client Write to  OSF1 Server : ~150Ko/s
>                                      ^^^^^^^
> 
> Do you know  if it's possible to have better results or if it's the maximun
> with the actual version of NFS Client for linux? 

This standard answer is right and wrong!

If you write tiny little blocks (e.g. GNU as) you get always a very low 
write performance (try iozone).

We had the same problems some times ago with HP-UX machines. Our solution
was in the exports man-page under export options:

           async     Specifying async increases write performance on the NFS
                     server by causing asynchronous writes on the NFS
                     server.  The async option can be specified anywhere on
                     the command line after the file system name.

Setting this option our write performance rises from 20-30K to 300-400K for all
blocksizes.
But the problem is, that i don't find this option in the OSF exports man-page.
The default setting for the OSF server seems to be snyc-write. But, there MUST
be a way set async writes! Please mail me if you solve this problem.

jh


------ Institut fuer Theoretische Physik, Universitaet Goettingen ------

 _______________________   ________.--'-`--._____   Tel:  +49 551 399363
|____==================_)  \_'===================`  FAX:  +49 551 399263
       _,--___.-|__|-.______|=====/  `---'          Post: Dr. J. A. Holm
       `---------._          ~~~~~|                       Bunsenstr. 9
                   `-._ -  -  - ,'                        37073 Goettingen
                       \_____,-'                          Deutschland
            finger: Holm@Theorie.Physik.UNI-Goettingen.DE
                        "Tee. Earl Grey. Heiss!"
Re: NFS Performance between Linux and OSF1
#732
Author: Eric Werme
Date: Tue, 02 Jul 1996 00:00
34 lines
2266 bytes 
WEILL Philippe <philippe.weill@aero.jussieu.fr> writes:

>When I use my OSF1 box as a server for my linux (i586) box 
>,performance is pretty bad

Standard problems with Digital UNIX NFS servers vs Linux clients include:

1) Adherence to the NFS V2 spec.
   Digital's systems don't respond to write requests until the data and metadata
   are on the disk.  HP and SGI don't do this and can lose data if the
   server crashes while NFS writes are being done.

2) "Write gathering" in the server.
   This is trying to get several writes to do them all at once.  This is great
   on Unix where the NFS clients tend to send several writes at once.  I think
   Linux doesn't do that, and a sleep delay in write gathering does you in.
   Try patching the kernel variable nfs_ufs_lbolt to be zero.  This will
   quicken the write replies and not have too great an impact on other
   Unix clients.

3) Buy a presto board for the server.
   This will have a big improvement.  It will also speed up local directory
   manipulation on the server (like speed up tar extraction, etc.)

4) Implement NFS V3 on Linux.  :-)
   V3 supports safe asynchronous writes and fixes the V2 write performance
   problem the right way.  Over FDDI, we can get V3 writes at 6 MB/sec
   without Presto (but with striped disks).

Ah - yes, 8KB transfers are the largest we support.
-- 
Eric (Ric) Werme         |  werme@zk3.dec.com
Digital Equipment Corp.  |  This space intentionally left blank.
Re: NFS Performance between Linux and OSF1
#730
Author: Andrew Gallatin
Date: Wed, 03 Jul 1996 00:00
26 lines
1856 bytes 
In article <werme.836329775@alingo.zk3.dec.com>,
Eric Werme <werme@zk3.dec.com> wrote:
>WEILL Philippe <philippe.weill@aero.jussieu.fr> writes:
>
>>When I use my OSF1 box as a server for my linux (i586) box 
>>,performance is pretty bad

<snip>
>4) Implement NFS V3 on Linux.  :-)
>   V3 supports safe asynchronous writes and fixes the V2 write performance
>   problem the right way.  Over FDDI, we can get V3 writes at 6 MB/sec
>   without Presto (but with striped disks).
>

Or switch from Linux to an OS that already has a working NFS V3
implementation ;-)  NetBSD-current and Solaris/x86 are two good
choices. Plus NetBSD will run your existing Linux binaries.

Drew

##############################################################################
#		Andrew Gallatin, Computer Project Manager                    #
#	    Institute of Statistics and Decision Sciences		     #
#	  Box 90251, Duke University, Durham, NC 27708-0251		     #
##############################################################################
Re: NFS Performance between Linux and OSF1
#731
Author: Wei Yang
Date: Wed, 03 Jul 1996 00:00
29 lines
1809 bytes 
Eric Werme wrote:
> 
> WEILL Philippe <philippe.weill@aero.jussieu.fr> writes:
> 
> >When I use my OSF1 box as a server for my linux (i586) box
> >,performance is pretty bad
> 

Hi, 

I also experienced similiar problem between my linux box and a DEC 
alpha. I used these two machines as both NFS servers and clients. Under
slackware 2.2, I added the number of nfsd and biod processes from 1 to 2
or 4 on linux and this could improve the proformance. 

Now I meet another problem. I am using HP workstation as NFS server and
a linux machine as a NFS client. and I am using slackware 3.0 which no
long has biod. Is there any similiar things function like boid or any
other method that I can use to improve the performance?


-- 
Wei Yang 
Dept. of Physics
Colorado State Univ.
Ft. Collins, CO 80523
Tel (970)-491-1581 (o)
wyang@omega.physics.colostate.edu
Re: NFS Performance between Linux and OSF1
#728
Author: Mark Swanson
Date: Wed, 10 Jul 1996 00:00
26 lines
1891 bytes 
On 3 Jul 1996 16:01:53 GMT, Andrew Gallatin <gallatin@davinci.isds.duke.edu> wrote:
><snip>
>>4) Implement NFS V3 on Linux.  :-)
>>   V3 supports safe asynchronous writes and fixes the V2 write performance
>>   problem the right way.  Over FDDI, we can get V3 writes at 6 MB/sec
>>   without Presto (but with striped disks).
>>
>
>Or switch from Linux to an OS that already has a working NFS V3
>implementation ;-)  NetBSD-current and Solaris/x86 are two good
>choices. Plus NetBSD will run your existing Linux binaries.
>
NetBSD doesn't run ELF executables, only a.out. a.out is obsolete and hasn't
been touched in a long time. The a.out libraries are unsupported.

-- 
***************************************************
Mark Swanson
BCSI                    fax:   613-580-7250
502-85 Range Rd.        http://www.bugless.com
Ottawa, Ontario, Canada email: mark@bugless.com
K1N 8J6
I will take freedom over government-compelled
responsibility in almost all cases, any time.
***************************************************
Re: NFS Performance between Linux and OSF1
#729
Author: Andrew Gallatin
Date: Wed, 10 Jul 1996 00:00
25 lines
1853 bytes 
In article <slrn4u681n.2t0.mark@hunter.mas.org>,
Mark Swanson <mark@bugless.com> wrote:
>On 3 Jul 1996 16:01:53 GMT, Andrew Gallatin <gallatin@davinci.isds.duke.edu> wrote:
>><snip>
>>>4) Implement NFS V3 on Linux.  :-)
>>>   V3 supports safe asynchronous writes and fixes the V2 write performance
>>>   problem the right way.  Over FDDI, we can get V3 writes at 6 MB/sec
>>>   without Presto (but with striped disks).
>>>
>>
>>Or switch from Linux to an OS that already has a working NFS V3
>>implementation ;-)  NetBSD-current and Solaris/x86 are two good
>>choices. Plus NetBSD will run your existing Linux binaries.
>>
>NetBSD doesn't run ELF executables, only a.out. a.out is obsolete and hasn't
>been touched in a long time. The a.out libraries are unsupported.

NetBSD-current (1.2Beta), does NFSv3 & NFSv2 and runs Linux elf &
a.out binaries.  You're thinking of NetBSD-1.1 which does only NFSv2
and runs Linux a.out binaries.

Cheers,

Drew
Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads