From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ob0-x232.google.com (mail-ob0-x232.google.com [IPv6:2607:f8b0:4003:c01::232]) (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 22C0421F38A for ; Thu, 7 May 2015 19:32:04 -0700 (PDT) Received: by obfe9 with SMTP id e9so45999992obf.1 for ; Thu, 07 May 2015 19:32:03 -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=bhI0KUvQdsRDMiYStEhoY8m4+5OwiZU8lFMYA97970s=; b=kQUtb5TtmcwWLWOaRGpd+kUUKLMY2QAtzzGZGz3/hgsMPSOI26xde+UkbRNVi0HHSL PD78VxjKtNDzJz5qev1+PHU0MV8xF+Z4ZTmgZRjlhOohXYyplJkHHpbhyizrSNYtiwuC muWnkgZ95q8F4Zz5ZBXOANs7APNVqU1hevYBpBZyFC4YUC7w6qkS9PvFKGw3aPRJujaU cLW+SBqzBbYC7pWSJ5wq7jNZLuQTwDy7lJDLfFAXOm3cfiuKzoy2BdeIxY0SfQzi9psE Q7q4GHLnspjITY3xJYXnOhGuH/6vffABGBdfkmopkZvTJb8t7pr9cxiZMyVusIibvYYU CwpQ== MIME-Version: 1.0 X-Received: by 10.182.233.134 with SMTP id tw6mr1287613obc.45.1431052323616; Thu, 07 May 2015 19:32:03 -0700 (PDT) Received: by 10.202.71.139 with HTTP; Thu, 7 May 2015 19:32:03 -0700 (PDT) Date: Thu, 7 May 2015 19:32:03 -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] packet mass, ecn, and a fractional count 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, 08 May 2015 02:32:33 -0000 from observing behaviors with large numbers of flows, fq_codel derived algorithms start to struggle to achieve the desired delay, cake seemingly less so, and perhaps this is related to collisions also. A thought would be to use a fractional (fixed point) increment to count rather than "1" when larger numbers of flows are present. The original ns2 model started applying a slightly different fraction to count in certain circumstances. It is quite doable in the 32 bit int to reserve a few bits for a fixed point fraction (say, 4 bits), but to apply it as the number of flows in play grows, not sure. See also the tracy widom distribution. --=20 Dave T=C3=A4ht Open Networking needs **Open Source Hardware** https://plus.google.com/u/0/+EricRaymond/posts/JqxCe2pFr67