public inbox for netdev@vger.kernel.org 
 help / color / mirror / Atom feed
From: lm@bitmover•com (Larry McVoy)
To: Linus Torvalds <torvalds@linux-foundation•org>
Cc: davem@davemloft•net, wscott@bitmover•com, netdev@vger•kernel.org
Subject: Re: tcp bw in 2.6
Date: Mon, 1 Oct 2007 17:59:18 -0700	[thread overview]
Message-ID: <20071002005917.GB5480@bitmover.com> (raw)
In-Reply-To: <alpine.LFD.0.999.0709291050200.3579@woody.linux-foundation.org>

On Sat, Sep 29, 2007 at 11:02:32AM -0700, Linus Torvalds wrote:
> On Sat, 29 Sep 2007, Larry McVoy wrote:
> > I haven't kept up on switch technology but in the past they were much
> > better than you are thinking.  The Kalpana switch that I had modified
> > to support vlans (invented by yours truly), did not store and forward,
> > it was cut through and could handle any load that was theoretically
> > possible within about 1%.
> 
> Hey, you may well be right. Maybe my assumptions about cutting corners are 
> just cynical and pessimistic. 

So I got a netgear switch and it works fine.  But my tests are busted.  
Catching netdev up, I'm trying to optimize traffic to a server that has
a gbit interface; I moved to a 24 port netgear that is all 10/100/1000
and I have a pile of clients to act as load generators.

I can do this on each of the clients 

	dd if=/dev/zero bs=1024000 | rsh work "dd of=/dev/null"

and that cranks up to about 47K packets/second which is about 70MB/sec.

One of my clients also has gigabit so I played around with just that
one and it (itanium running hpux w/ broadcom gigabit) can push the load
as well.  One weird thing is that it is dependent on the direction the
data is flowing.  If the hp is sending then I get 46MB/sec, if linux is
sending then I get 18MB/sec.  Weird.  Linux is debian, running 

    Linux work 2.6.18-5-k7 #1 SMP Thu Aug 30 02:52:31 UTC 2007 i686 

and dual e1000 cards:

    e1000: eth0: e1000_probe: Intel(R) PRO/1000 Network Connection
    e1000: eth1: e1000_probe: Intel(R) PRO/1000 Network Connection

I wrote a tiny little program to try and emulate this and I can't get
it to do as well.  I've tracked it down, I think, to the read side.
The server sources, the client sinks, the server looks like:

11689 accept(3, {sa_family=AF_INET, sin_port=htons(49376), sin_addr=inet_addr("10.3.1.38")}, [16]) = 4
11689 setsockopt(4, SOL_SOCKET, SO_RCVBUF, [1048576], 4) = 0
11689 setsockopt(4, SOL_SOCKET, SO_SNDBUF, [1048576], 4) = 0
11689 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xb7ddf708) = 11694
11689 close(4)                          = 0
11689 accept(3,  <unfinished ...>
11694 write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 1048576
11694 write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 1048576
11694 write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 1048576
11694 write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 1048576
...

but the client looks like

connect(3, {sa_family=AF_INET, sin_port=htons(31235), sin_addr=inet_addr("10.3.9.1")}, 16) = 0
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 1448
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 1448
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1048576) = 2896

which I suspect may be the problem.

I played around with SO_RCVBUF/SO_SNDBUF and that didn't help.  So any ideas why
a simple dd piped through rsh is kicking my ass?  It must be something simple
but my test program is tiny and does nothing weird that I can see.
-- 
---
Larry McVoy                lm at bitmover.com           http://www.bitkeeper.com

       reply	other threads:[~2007-10-02  1:30 UTC|newest]

Thread overview: 56+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20070929142517.EC6AB5FB21@work.bitmover.com>
     [not found] ` <alpine.LFD.0.999.0709290914410.3579@woody.linux-foundation.org>
     [not found]   ` <20070929172639.GB7037@bitmover.com>
     [not found]     ` <alpine.LFD.0.999.0709291050200.3579@woody.linux-foundation.org>
2007-10-02  0:59       ` Larry McVoy [this message]
2007-10-02  2:14         ` tcp bw in 2.6 Linus Torvalds
2007-10-02  2:20           ` Larry McVoy
2007-10-02  3:50             ` David Miller
2007-10-02  4:23               ` Larry McVoy
2007-10-02 15:06             ` John Heffner
2007-10-02 17:14             ` Rick Jones
2007-10-02 17:20               ` Larry McVoy
2007-10-02 18:01                 ` Rick Jones
2007-10-02 18:40                   ` Larry McVoy
2007-10-02 19:47                     ` Rick Jones
2007-10-02 21:32                     ` David Miller
2007-10-03  7:19               ` Bill Fink
2007-10-02 10:52         ` Herbert Xu
2007-10-02 15:09           ` Larry McVoy
2007-10-02 15:41             ` Larry McVoy
2007-10-02 16:25               ` Larry McVoy
2007-10-02 16:47                 ` Stephen Hemminger
2007-10-02 16:49                   ` Larry McVoy
2007-10-02 17:10                     ` Stephen Hemminger
2007-10-15 12:40                   ` Daniel Schaffrath
2007-10-15 15:49                     ` Stephen Hemminger
2007-10-02 16:34               ` Linus Torvalds
2007-10-02 16:48                 ` Larry McVoy
2007-10-02 21:16                   ` David Miller
2007-10-02 21:26                     ` Larry McVoy
2007-10-02 21:47                       ` David Miller
2007-10-02 22:17                         ` Rick Jones
2007-10-02 22:32                           ` David Miller
2007-10-02 22:36                             ` Larry McVoy
2007-10-02 22:59                               ` Rick Jones
2007-10-03  8:02                               ` David Miller
2007-10-02 16:48               ` Ben Greear
2007-10-02 17:11                 ` Larry McVoy
2007-10-02 17:18                   ` Ben Greear
2007-10-02 17:21                     ` Larry McVoy
2007-10-02 17:54                       ` Stephen Hemminger
2007-10-02 18:35                         ` Larry McVoy
2007-10-02 18:29             ` John Heffner
2007-10-02 19:07               ` Larry McVoy
2007-10-02 19:29                 ` Linus Torvalds
2007-10-02 20:31                   ` David Miller
2007-10-02 19:33                 ` Larry McVoy
2007-10-02 19:53                   ` John Heffner
2007-10-02 20:14                     ` Larry McVoy
2007-10-02 20:40                       ` Rick Jones
2007-10-02 20:42                       ` Wayne Scott
2007-10-02 21:56                         ` Linus Torvalds
2007-10-02 19:27             ` Linus Torvalds
2007-10-02 19:53               ` Rick Jones
2007-10-02 20:33               ` David Miller
2007-10-02 20:44                 ` Roland Dreier
2007-10-02 21:21                 ` Larry McVoy
2007-10-03 21:13                   ` Pekka Pietikainen
2007-10-03 21:23                     ` Larry McVoy
2007-10-03 21:50                       ` Pekka Pietikainen

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=20071002005917.GB5480@bitmover.com \
    --to=lm@bitmover$(echo .)com \
    --cc=davem@davemloft$(echo .)net \
    --cc=netdev@vger$(echo .)kernel.org \
    --cc=torvalds@linux-foundation$(echo .)org \
    --cc=wscott@bitmover$(echo .)com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html
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