From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH v2 1/1] TCP: increase default initial receive window. Date: Tue, 21 Dec 2010 06:00:06 +0100 Message-ID: <1292907606.2627.151.camel@edumazet-laptop> References: <1292642451-892-1-git-send-email-nanditad@google.com> <1292890556-29904-1-git-send-email-nanditad@google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: "David S. Miller" , netdev@vger.kernel.org, Tom Herbert , Laurent Chavey , Yuchung Cheng , Stephen Hemminger , Rick Jones To: Nandita Dukkipati Return-path: Received: from mail-ww0-f44.google.com ([74.125.82.44]:35236 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750773Ab0LUFAM (ORCPT ); Tue, 21 Dec 2010 00:00:12 -0500 Received: by wwa36 with SMTP id 36so3754624wwa.1 for ; Mon, 20 Dec 2010 21:00:11 -0800 (PST) In-Reply-To: <1292890556-29904-1-git-send-email-nanditad@google.com> Sender: netdev-owner@vger.kernel.org List-ID: Le lundi 20 d=C3=A9cembre 2010 =C3=A0 16:15 -0800, Nandita Dukkipati a = =C3=A9crit : > This patch changes the default initial receive window to 10 mss > (defined constant). The default window is limited to the maximum > of 10*1460 and 2*mss (when mss > 1460). >=20 > draft-ietf-tcpm-initcwnd-00 is a proposal to the IETF that recommends > increasing TCP's initial congestion window to 10 mss or about 15KB. > Leading up to this proposal were several large-scale live Internet > experiments with an initial congestion window of 10 mss (IW10), where > we showed that the average latency of HTTP responses improved by > approximately 10%. This was accompanied by a slight increase in > retransmission rate (0.5%), most of which is coming from applications > opening multiple simultaneous connections. To understand the extreme > worst case scenarios, and fairness issues (IW10 versus IW3), we furth= er > conducted controlled testbed experiments. We came away finding minima= l > negative impact even under low link bandwidths (dial-ups) and small > buffers. These results are extremely encouraging to adopting IW10. >=20 > However, an initial congestion window of 10 mss is useless unless a T= CP > receiver advertises an initial receive window of at least 10 mss. > Fortunately, in the large-scale Internet experiments we found that mo= st > widely used operating systems advertised large initial receive window= s > of 64KB, allowing us to experiment with a wide range of initial > congestion windows. Linux systems were among the few exceptions that > advertised a small receive window of 6KB. The purpose of this patch i= s > to fix this shortcoming. >=20 > References: > 1. A comprehensive list of all IW10 references to date. > http://code.google.com/speed/protocols/tcpm-IW10.html >=20 > 2. Paper describing results from large-scale Internet experiments wit= h IW10. > http://ccr.sigcomm.org/drupal/?q=3Dnode/621 >=20 > 3. Controlled testbed experiments under worst case scenarios and a > fairness study. > http://www.ietf.org/proceedings/79/slides/tcpm-0.pdf >=20 > 4. Raw test data from testbed experiments (Linux senders/receivers) > with initial congestion and receive windows of both 10 mss. > http://research.csc.ncsu.edu/netsrv/?q=3Dcontent/iw10 >=20 > 5. Internet-Draft. Increasing TCP's Initial Window. > https://datatracker.ietf.org/doc/draft-ietf-tcpm-initcwnd/ >=20 > Signed-off-by: Nandita Dukkipati Acked-by: Eric Dumazet