From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ob0-x22c.google.com (mail-ob0-x22c.google.com [IPv6:2607:f8b0:4003:c01::22c]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority G2" (verified OK)) by huchra.bufferbloat.net (Postfix) with ESMTPS id 6FE3621F353 for ; Thu, 11 Jun 2015 18:09:00 -0700 (PDT) Received: by obcej4 with SMTP id ej4so13929801obc.0 for ; Thu, 11 Jun 2015 18:09:00 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=XIfjJtxrBFXv8f8ipN++6gXKT9eFrya8PtBtHaPLAVk=; b=sq3gMNL496TaYaBs//bznvIrVbRRkHRKCoTSdZmgkLi7JfQelbOXOKhFFVkPufRX7e iJfblmgunXp95TyXHBN9Ww+4gEHBFTHwC7a9ENPpZx/GZUQYl/560FbbwV0evTUOVsz9 2vmE2t9d4vOqA1rvuI4BNjqoZ9TG4l/7glB3FsNPHuVHeBEjTACc/7lQ2/bih2axY0uH KlwyeYxfkDOInaYyM4ne5kPBFq8VEVfW9KmUS/1QpMIRY/0Qg0QL4ll7L4mqzMGrgR9R H02/CP3psJIDYIwIbFg1/+hSkioJnIXqBgdxEqVrKUhx701AsieLCMRVDuJAcgsgeZkT 9ewg== MIME-Version: 1.0 X-Received: by 10.182.240.135 with SMTP id wa7mr10088532obc.63.1434071339840; Thu, 11 Jun 2015 18:08:59 -0700 (PDT) Received: by 10.202.105.129 with HTTP; Thu, 11 Jun 2015 18:08:59 -0700 (PDT) Date: Thu, 11 Jun 2015 18:08:59 -0700 Message-ID: From: Dave Taht To: cake@lists.bufferbloat.net Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Subject: [Cake] an "evil" queue class? X-BeenThere: cake@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: Cake - FQ_codel the next generation List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jun 2015 01:09:29 -0000 An evil flow is one that is consistently not responding to drops, and occupying a vastly disproportionate amount of queue space compared to the other flows. The HHF qdisc has a concept like this in it, but what I have in mind is simpler. Somewhat related, we can make cake_drop be less cpu intensive and more deterministic with a small cache of the top (say, 3) "offenders" maintained at either dequeue or enqueue time. (ideas based on pounding codel flat with iperf -u and watching it stay pounded flat for a long period of time) --=20 Dave T=C3=A4ht What will it take to vastly improve wifi for everyone? https://plus.google.com/u/0/explore/makewififast