public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Thiemo Voigt <thiemo@sics.se>
To: kuznet@ms2.inr.ac.ru
Cc: Sridhar Samudrala <samudrala@us.ibm.com>,
	alan@lxorguk.ukuu.org.uk, linux-kernel@vger.kernel.org,
	linux-net@vger.kernel.org, lartc@mailman.ds9a.nl,
	diffserv-general@lists.sourceforge.net, rusty@rustcorp.com.au
Subject: Re: [PATCH] Inbound Connection Control mechanism: Prioritized Accept
Date: Sat, 28 Jul 2001 22:01:04 +0200	[thread overview]
Message-ID: <3B631A00.8E860DC1@sics.se> (raw)
In-Reply-To: <200107281912.XAA17362@ms2.inr.ac.ru>

kuznet@ms2.inr.ac.ru wrote:

> Hello!
>
> > Low priority connections can clog the accept queue only when there are no
> > high priority connection requests coming along. As soon as a slot becomes empty
> > in the accept queue, it becomes available for a high priority connection.
>
> And in presence of persistent low priority traffic, high priority connection
> will not have any chances to take this slot. When high priority connection
> arrives all the slots are permanently busy with low ones.
>
> > If that happens, TCP SYN policing can be employed to limit the rate of low
> > priority connections getting into accept queue.
>

The aim of TCP SYN policing is to prevent server overload by discarding
connection requests early when the server system is about to reach overload.
One of the indicators of overload might be that the accept queue is
close to being filled up, there is little CPU time etc.
In these cases, TCP SYN policing should adapt (i.e. lower) the acceptance rates.
In such an adaptive  system, the accept queue is not supposed to be completely
filled, thus low priority connections are not able to starve high priority
connections. By the way, different acceptance rates can be given
to different priority classes.

A more detailed discussion than on the website can be found
in the paper "In-kernel mechanisms for adaptive  control of
overloaded web servers", available at
http://wwwtgs.cs.utwente.nl/Docs/eunice/summerschool/papers/programme.html
This paper discusses TCP SYN policing and prioritized listen queue.


Cheers,
Thiemo



  reply	other threads:[~2001-07-28 20:01 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2001-07-27 17:10 [PATCH] Inbound Connection Control mechanism: Prioritized Accept Queue Sridhar Samudrala
2001-07-27 17:25 ` [PATCH] Inbound Connection Control mechanism: Prioritized Accept Alan Cox
2001-07-27 18:01   ` Sridhar Samudrala
2001-07-27 18:07     ` Alan Cox
2001-07-27 18:04   ` kuznet
2001-07-27 19:55     ` Sridhar Samudrala
2001-07-28 19:12       ` kuznet
2001-07-28 20:01         ` Thiemo Voigt [this message]
2001-07-29 16:25           ` kuznet
2001-07-30  7:40             ` Sridhar Samudrala
2001-08-01 23:22               ` Alexey Kuznetsov
2001-08-02 17:17                 ` Sridhar Samudrala
2001-08-02 22:31                   ` Alexey Kuznetsov

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=3B631A00.8E860DC1@sics.se \
    --to=thiemo@sics.se \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=diffserv-general@lists.sourceforge.net \
    --cc=kuznet@ms2.inr.ac.ru \
    --cc=lartc@mailman.ds9a.nl \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-net@vger.kernel.org \
    --cc=rusty@rustcorp.com.au \
    --cc=samudrala@us.ibm.com \
    /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