Re: [aqm] [tcpm] TCP ACK Suppression

David Lang <david@lang.hm> Sat, 10 October 2015 03:30 UTC

Return-Path: <david@lang.hm>
X-Original-To: aqm@ietfa.amsl.com
Delivered-To: aqm@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id AFAEF1B54C5; Fri, 9 Oct 2015 20:30:09 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.91
X-Spam-Level:
X-Spam-Status: No, score=-1.91 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, T_RP_MATCHES_RCVD=-0.01] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id R2AfAkgToTUM; Fri, 9 Oct 2015 20:30:08 -0700 (PDT)
Received: from bifrost.lang.hm (mail.lang.hm [64.81.33.126]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 260BF1B54C0; Fri, 9 Oct 2015 20:30:07 -0700 (PDT)
Received: from asgard.lang.hm (asgard.lang.hm [10.0.0.100]) by bifrost.lang.hm (8.13.4/8.13.4/Debian-3) with ESMTP id t9A3TtuT026712; Fri, 9 Oct 2015 20:29:55 -0700
Date: Fri, 9 Oct 2015 20:29:55 -0700 (PDT)
From: David Lang <david@lang.hm>
X-X-Sender: dlang@asgard.lang.hm
To: Joe Touch <touch@isi.edu>
In-Reply-To: <alpine.DEB.2.02.1510091910170.3717@nftneq.ynat.uz>
Message-ID: <alpine.DEB.2.02.1510092019100.15683@nftneq.ynat.uz>
References: <5618005A.8070303@isi.edu> <70335.1444421059@lawyers.icir.org> <D23D8CA5.54DF5%g.white@cablelabs.com> <56183B49.4000506@isi.edu> <alpine.DEB.2.02.1510091511540.3717@nftneq.ynat.uz> <56183E93.1010308@isi.edu> <alpine.DEB.2.02.1510091528320.3717@nftneq.ynat.uz> <5618420E.9040609@isi.edu> <alpine.DEB.2.02.1510091628010.3717@nftneq.ynat.uz> <5618554F.3080103@isi.edu> <alpine.DEB.2.02.1510091716100.3717@nftneq.ynat.uz> <56185E44.9050702@isi.edu> <alpine.DEB.2.02.1510091910170.3717@nftneq.ynat.uz>
User-Agent: Alpine 2.02 (DEB 1266 2009-07-14)
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
Archived-At: <http://mailarchive.ietf.org/arch/msg/aqm/JN6FwT7FahtlaFLlEUAVgvupQW8>
Cc: "LAUTENSCHLAEGER, Wolfram \(Wolfram\)" <wolfram.lautenschlaeger@alcatel-lucent.com>, Greg White <g.white@CableLabs.com>, "tcpm@ietf.org" <tcpm@ietf.org>, "aqm@ietf.org" <aqm@ietf.org>, "mallman@icir.org" <mallman@icir.org>
Subject: Re: [aqm] [tcpm] TCP ACK Suppression
X-BeenThere: aqm@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: "Discussion list for active queue management and flow isolation." <aqm.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/aqm>, <mailto:aqm-request@ietf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/aqm/>
List-Post: <mailto:aqm@ietf.org>
List-Help: <mailto:aqm-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/aqm>, <mailto:aqm-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 10 Oct 2015 03:30:09 -0000

On Fri, 9 Oct 2015, David Lang wrote:

> On Fri, 9 Oct 2015, Joe Touch wrote:
>
>> On 10/9/2015 5:22 PM, David Lang wrote:
>>> You don't want to acknowlege it, but TCP is broken in the face of
>>> excessive buffering.
>> 
>> Arguably, buffering was broken and failed to provide the feedback to TCP
>> (see next paragraph)
>> 
>>> TCPM isn't fixing that, grassroots efforts are
>>> developing the fixes and AQM is formalizing the results.
>> 
>> TCPM fixed it in 2001 by providing the flags for ECN, which was enabled
>> by default in Windows since 2012. ALTQ support for ECN has been around
>> for nearly that long.
>> 
>> What's changed? Not the TCP reaction (except in extreme cases such as
>> for datacenters) but the router algs. And getting the router algs into
>> routers - esp. home devices.
>
> if ECN solved the problem, then bufferbloat would have been a non-issue. A 
> large percentage of the home routers are running OpenWRT or similar with ECN 
> enabled. That didn't solve the problem.

when the investigation into bufferbloat started, we were assuming that the fix 
was just to reduce the size of the buffers and/or make sure that ECN was 
enabled. Experimentation showed that there was no buffer size that both enabled 
full utilization of the links and avoided excessive delays. Switching buffers 
from X packets to X bytes helped a lot, but not enough.

Part of what's changed is the usage patterns, but a lot of it is just that 
things that worked when a fast network was 10Mb don't always scale across 
several orders of magnatude.

David Lang