netdev - Re: [PATCH] tcp: avoid cwnd moderation in undo

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Wed, 16 Mar 2011 22:24:42 +0200 (EET)
From:	"Ilpo Järvinen" <ilpo.jarvinen@...sinki.fi>
To:	John Heffner <johnwheffner@...il.com>
cc:	Carsten Wolff <carsten@...ffcarsten.de>,
	Yuchung Cheng <ycheng@...gle.com>,
	David Miller <davem@...emloft.net>,
	Nandita Dukkipati <nanditad@...gle.com>,
	Netdev <netdev@...r.kernel.org>,
	Alexander Zimmermann <alexander.zimmermann@...sys.rwth-aachen.de>
Subject: Re: [PATCH] tcp: avoid cwnd moderation in undo

On Wed, 16 Mar 2011, John Heffner wrote:

> On Wed, Mar 16, 2011 at 12:18 PM, Carsten Wolff <carsten@...ffcarsten.de> wrote:
> > Unfortunately, no. ;-) My point is, that cwnd should be moderated when the
> > congestion state changes are undone after a spurious recovery has been
> > detected. Reordering is only one possible reason for a false recovery. And I
> > stick to that point because of the thoughts I pointed out in my mail to john,
> > i.e. undo typically leading to exceptionally large segment bursts.
> 
> There's a more general discussion here as to how much it's worth
> avoiding bursts at all.  What research on the subject I'm aware of is
> somewhat inconclusive.  It's possible to construct scenarios where
> bursting, burst suppression, or pacing each win or lose badly.  If you
> can choose only one approach as one-size-fits all it's difficult with
> the information at hand to pick only one.  However, I really do wonder
> why it's so important to suppress bursts on congestion state undo when
> other, likely far more common, sources of bursts are not suppressed.
> 
> Carsten, do you have any specific examples of cases you're concerned
> about?  FWIW, there are exactly two causes for spurious retransmits:
> spurious fast retransmit due to reordering, and spurious timeouts due
> to a delay spike.  Are you particularly concerned with one more than
> the other?

The latter is handled with FRTO (defaults on in Linux) so nicely that no 
burstiness can be seen.

-- 
 i.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html