netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ben Hutchings <bhutchings@solarflare.com>
To: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Neal Cardwell <ncardwell@google.com>,
	Netdev <netdev@vger.kernel.org>, <alekcejk@googlemail.com>
Subject: Re: limited network bandwidth with 3.2.x kernels
Date: Thu, 16 Feb 2012 13:51:05 +0000	[thread overview]
Message-ID: <1329400265.3048.178.camel@deadeye> (raw)
In-Reply-To: <1329399616.9395.9.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC>

On Thu, 2012-02-16 at 14:40 +0100, Eric Dumazet wrote:
> Le jeudi 16 février 2012 à 08:29 +0100, Eric Dumazet a écrit :
> > So I took the time to setup a netem in my lab (and had to fix netem by
> > the way). Gigabit link.
> > 
> > On sender :
> > tc qdisc add dev vlan.103 root netem delay 50ms
> > 
> > And netperf session is a bit strange, since receiver window is about
> > 1Mbytes (17097*64) but sender never uses half of it.
> > 
> > cwnd is ~408
> > 
> > $ ss -emoi  dst 192.168.20.108
> > State      Recv-Q Send-Q                                 Local
> > Address:Port                                     Peer Address:Port   
> > ESTAB      0      450976
> > 192.168.20.110:52017
> > 192.168.20.108:44169    timer:(on,235ms,0) ino:21711 sk:f24d2d00
> > 	 mem:(r0,w698880,f206336,t0) ts sack ecn bic wscale:6,8 rto:252
> > rtt:52/0.75 cwnd:408 send 90.9Mbps rcv_space:14600
> > 
> > $ tc -s -d qdisc show dev vlan.103
> > qdisc netem 8009: root refcnt 2 limit 1000 delay 50.0ms
> >  Sent 15149167430 bytes 10017438 pkt (dropped 0, overlimits 0 requeues
> > 0) 
> >  rate 74878Kbit 6193pps backlog 476760b 316p requeues 0 
> 
> Its seems a problem with GRO.
> 
> Receiver opens its window each time it sends an ACK, without taking care
> how many segments were ACKed.
> 
> In these traces, I give the ACKS sent by receiver, first trace GRO on,
> second with GRO off
> 
> We can see that with GRO off, receiver opens its window way faster
> (against number of received bytes)
> 
> (REC & SENDER are both 3.0 kernels, so problem is quite old)
[...]

I'm aware of this problem and I believe it exists with most
implementations of LRO.

The out-of-tree version of the sfc driver has its own soft-LRO
implementation (SSR) which does some limited connection tracking to
detect slow start and disable aggregation temporarily.  I've been
meaning to look into enhancing GRO to match SSR, but haven't got round
to it yet.

Ben.

-- 
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.

  reply	other threads:[~2012-02-16 13:51 UTC|newest]

Thread overview: 66+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-02-13 20:04 limited network bandwidth with 3.2.x kernels alekcejk
2012-02-13 20:18 ` Eric Dumazet
2012-02-13 21:13   ` Eric Dumazet
2012-02-13 21:58     ` alekcejk
2012-02-13 22:44       ` Eric Dumazet
2012-02-13 22:51         ` Neal Cardwell
2012-02-13 23:09           ` alekcejk
2012-02-13 23:14             ` Eric Dumazet
2012-02-13 23:24               ` alekcejk
2012-02-13 23:49         ` alekcejk
2012-02-14  3:06           ` Neal Cardwell
2012-02-14  3:26             ` alekcejk
2012-02-14  5:55           ` Eric Dumazet
2012-02-14 14:09             ` alekcejk
2012-02-13 21:16   ` alekcejk
2012-02-21  4:21   ` alekcejk
2012-02-21 16:45     ` Eric Dumazet
2012-02-21 17:19       ` alekcejk
2012-02-21 17:31         ` Eric Dumazet
2012-02-21 18:53           ` alekcejk
2012-02-22  1:26           ` alekcejk
     [not found] ` <1920135.kl3o7Tt79c@localhost.localdomain>
     [not found]   ` <1329228798.4818.0.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC>
2012-02-14 14:28     ` alekcejk
2012-02-14 16:00       ` Eric Dumazet
2012-02-14 16:43         ` alekcejk
2012-02-15  4:43           ` Eric Dumazet
2012-02-15  6:03             ` Neal Cardwell
2012-02-15  6:06               ` Eric Dumazet
2012-02-15  7:52                 ` Bill Fink
2012-02-15  8:00                   ` Eric Dumazet
2012-02-15  8:18                     ` Bill Fink
2012-02-15  8:31                       ` Eric Dumazet
2012-02-15 13:44                   ` alekcejk
2012-02-15 13:43               ` Re: Re: Re: Re: " alekcejk
2012-02-15 13:44             ` alekcejk
2012-02-15 13:49               ` Eric Dumazet
2012-02-15 13:53                 ` Eric Dumazet
2012-02-15 14:11                 ` alekcejk
2012-02-15 14:49                   ` Neal Cardwell
2012-02-15 14:56                     ` Eric Dumazet
2012-02-15 15:05                     ` alekcejk
2012-02-15 15:08                   ` Eric Dumazet
2012-02-15 19:44                     ` Neal Cardwell
2012-02-16  7:29                       ` Eric Dumazet
2012-02-16 13:40                         ` Eric Dumazet
2012-02-16 13:51                           ` Ben Hutchings [this message]
2012-02-16 15:09                             ` Eric Dumazet
2012-02-16 16:37                               ` Ben Hutchings
2012-02-16 17:01                               ` David Miller
2012-02-16 17:22                                 ` Neal Cardwell
2012-02-16 17:54                                   ` alekcejk
2012-02-16 18:19                                     ` Neal Cardwell
2012-02-16 21:49                                       ` alekcejk
2012-02-16 21:59                                         ` Eric Dumazet
2012-02-16 22:08                                           ` alekcejk
2012-02-16 22:35                                             ` Eric Dumazet
2012-02-17 16:41                                               ` Neal Cardwell
2012-02-21 18:41                                                 ` Eric Dumazet
2012-02-21 19:11                                                   ` David Miller
2012-02-22  5:51                                                   ` Neal Cardwell
2012-02-22  7:36                                                     ` Eric Dumazet
2012-02-23 18:39                                                       ` Neal Cardwell
2012-02-23 18:49                                                         ` Eric Dumazet
2012-02-27 19:39                                                         ` David Miller
2012-02-17 16:55                                           ` Re: " Neal Cardwell
2012-02-16 18:22                                   ` Eric Dumazet
2012-02-16 19:44                                     ` Eric Dumazet

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=1329400265.3048.178.camel@deadeye \
    --to=bhutchings@solarflare.com \
    --cc=alekcejk@googlemail.com \
    --cc=eric.dumazet@gmail.com \
    --cc=ncardwell@google.com \
    --cc=netdev@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;
as well as URLs for NNTP newsgroup(s).