All of lore.kernel.org
 help / color / mirror / Atom feed
From: Andy Furniss <andy.furniss@dsl.pipex.com>
To: lartc@vger.kernel.org
Subject: Re: [LARTC] Please advise how to set up for 5 IPs
Date: Thu, 30 Sep 2004 12:46:34 +0000	[thread overview]
Message-ID: <415C002A.1050501@dsl.pipex.com> (raw)
In-Reply-To: <4159BD3F.45CB63E1@iswest.com>

Jason Boxman wrote:
> On Tuesday 28 September 2004 15:36, gypsy wrote:
> <snip>
> 
>>How can I implement ingress shaping / policing to limit the combined
>>incoming rate, regardless of IP, to about 700K (the connection is
>>actually 730K in and 690K out).  Only because it REALLY annoys me to
>>drop an already received packet, I would prefer to shape than to drop,
>>but I am presently policing.
> 
> 
> I can't see how you could shape other than to drop.  You have already received 
> the packet.  If it's over the rate for that particular flow, and the traffic 
> is TCP, you can only hope enough drops will result in the connection being 
> throttled by the remote end's TCP stack.

Queueing is important aswell as dropping. You will need to drop 
sometimes, but you can decide to only queue/drop bulk traffic if you 
shape with htb. For ingress, or more correctly, shaping from the wrong 
and of a bottleneck I like to use esfq (so I can choose length) with 
classic hash - so that each tcp gets dequeued at the rate it's entitled 
- rather than a burst at near full speed followed by a wait while others 
are serviced. Shaping from the narrow end of the bottleneck is hard if 
you care alot about latency and don't want to sacrifice too much 
bandwidth. SFQ helps prevent burstiness over other queues - but don't 
set perturb too low - Default (10) is OK, I use 20. It causes packet 
reordering which hurts latency when shaping incoming - the effect is 
worse with single connections.

There are other tweaks possible but they are probably not worth the 
hassle unless you are noticing problems.

Andy.


_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

      parent reply	other threads:[~2004-09-30 12:46 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-09-28 19:36 [LARTC] Please advise how to set up for 5 IPs gypsy
2004-09-28 19:41 ` Jason Boxman
2004-09-30 12:46 ` Andy Furniss [this message]

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=415C002A.1050501@dsl.pipex.com \
    --to=andy.furniss@dsl.pipex.com \
    --cc=lartc@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 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.