[tcpm] (reducing) tcp bursts

Yuchung Cheng <ycheng@google.com> Tue, 30 July 2013 11:20 UTC

Return-Path: <ycheng@google.com>
X-Original-To: tcpm@ietfa.amsl.com
Delivered-To: tcpm@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1DC1311E814F for <tcpm@ietfa.amsl.com>; Tue, 30 Jul 2013 04:20:58 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.367
X-Spam-Level:
X-Spam-Status: No, score=-1.367 tagged_above=-999 required=5 tests=[AWL=0.011, BAYES_00=-2.599, FM_FORGED_GMAIL=0.622, J_CHICKENPOX_33=0.6, NO_RELAYS=-0.001]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id YzcwB+92t+aI for <tcpm@ietfa.amsl.com>; Tue, 30 Jul 2013 04:20:57 -0700 (PDT)
Received: from mail-oa0-x231.google.com (mail-oa0-x231.google.com [IPv6:2607:f8b0:4003:c02::231]) by ietfa.amsl.com (Postfix) with ESMTP id 3367421F8994 for <tcpm@ietf.org>; Tue, 30 Jul 2013 04:20:56 -0700 (PDT)
Received: by mail-oa0-f49.google.com with SMTP id n16so4241078oag.8 for <tcpm@ietf.org>; Tue, 30 Jul 2013 04:20:55 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:from:date:message-id:subject:to:content-type; bh=gcQoReSB7fRZjGabfNMCIUhsLQt9rtPow6eyAUg93DE=; b=L8MxdB2hfJ7YHoktVfnQa+nHtALM2Jqq35alg0T15hHCFnAuD2wvmAJTtIBwTuRUcp aYLWSy5qkCeWF7Kn54rUiaFYCRGvwn7bPusLnpR4cs98By7hOxC+eK+UYTU4Cm1BjZwr sHSXqv6q3HE2cCt649a2KdvQcQ5FVXq3iZwPUrMO1hJEp1w30+U+RAydTKEYNk5ca7FP JrcQqtGUrE7F2K4hXowpsshEH6BsdZlX0GyN2UzkbqxQEHUoBd+ds4yKKXGvXa4uMO+n 5KhdeeGw8Uh1dxMv62f2/YkeNjDtwcfLnJFeNoV03Cqw2FxRlp5AymdfdnWTh/TSdJ41 Q8lg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:from:date:message-id:subject:to:content-type :x-gm-message-state; bh=gcQoReSB7fRZjGabfNMCIUhsLQt9rtPow6eyAUg93DE=; b=b3A/sAa0ml49u6j0gToBeckIWIpPtoSBLuw/iV5111IuJxuvJkIyHDhSlY5wBE8ddW LnwJNFH/NKt14wN/OMdIrnv9IZsX4vqqCqLXAJwPar3f17erqkof3ffTWXsJEbJScnk4 C1Y+gdkaVygNNc7Wj78OF7rrC2cPnLPF/AhkKgiav/33CVuY1/ZaVFCWGdKBpnslMzfo 6hD3I4iwufJhTGAwcwaeNXlhGkxEHidKTSBlVQI33J8EdLA/f0b9FR8jifflCN4w3vzP b5HqUB5Sr236n5hq9Q8T3O4PYdnSRvgFbeSGjZLerOlIYG4XLcIUTmbMiEoVMmNmkeMg tvqQ==
X-Received: by 10.42.173.136 with SMTP id r8mr802498icz.26.1375183255241; Tue, 30 Jul 2013 04:20:55 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.21.71 with HTTP; Tue, 30 Jul 2013 04:20:34 -0700 (PDT)
From: Yuchung Cheng <ycheng@google.com>
Date: Tue, 30 Jul 2013 13:20:34 +0200
Message-ID: <CAK6E8=cCyFg7YtDnJy=NLcoCtQJX+S6J65fBUYCrGqaxwHUJDw@mail.gmail.com>
To: "tcpm@ietf.org Extensions" <tcpm@ietf.org>
Content-Type: text/plain; charset="ISO-8859-1"
X-Gm-Message-State: ALoCoQnbw7y4UComezVrOVPaEG2HgNVdy5jSEg29oRVVePk/HA2pD2pVtMs9QXypRxzp9bXgGnmbsDAjGJ2Q3HkM9FKuoNur2tFctFDtOgWZgQTKpoCWj7MFPAyJX052aKyN2DUPhK0Qm01VdS1AbGQx4uo4Y5ZFKsOvLbmr2aVg0ubvy3ru/ydP1fJ2+csxmjtjXMW2RHG/
Subject: [tcpm] (reducing) tcp bursts
X-BeenThere: tcpm@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: TCP Maintenance and Minor Extensions Working Group <tcpm.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tcpm>, <mailto:tcpm-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tcpm>
List-Post: <mailto:tcpm@ietf.org>
List-Help: <mailto:tcpm-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tcpm>, <mailto:tcpm-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 30 Jul 2013 11:20:58 -0000

There are a lot of discussion on bursts across talks.
1. newcwv: idle-restart
2. tlp: how often is tail drops be caused by (higher) initial burst/send
3. burst (loss) after recovery due to snd.una + rwin jump.
4. I can throw in another one: video player application throttle
sender by not reading the socket or clamp the receive buffer. But this
causes TCP to burst when rwin opens up.

I think the working group should work on a general solution to reduce
burst in the window-based, ack-clocked, TCP. I have heard solutions
like
1. BSD/randy's max-burst solution
2. pace cwnd/rtt but in max-burst chunks
3. more ideas in http://www.isi.edu/touch/pubs/draft-hughes-restart-00.txt

We all know TCP is very smooth in bulk transfer. Unfortunately modern
Apps are chatty even on video.

Thoughts?