From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-wi0-x233.google.com (mail-wi0-x233.google.com [IPv6:2a00:1450:400c:c05::233]) (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 32E9D21F113 for ; Fri, 20 Dec 2013 18:33:27 -0800 (PST) Received: by mail-wi0-f179.google.com with SMTP id z2so4461145wiv.0 for ; Fri, 20 Dec 2013 18:33:26 -0800 (PST) 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=2PZ0jb4bYTaaLvU8JXezTqTaKu2aPbg6C8eKoYsNjuc=; b=E69rPHlmSUQfSL+FVvbqhQdpc+ucX1zIGlYV4gq/Q3g3HME1+R1qYkKLCAf4EFlrjh OnShX57NzJKRTu38J5mwqqCFUrA24Gd9XaFyRpQ4JCG9mTxJ6vfEL8maGhqNXqBR+6Jp Be9JemaU/Yv622t0CThy3alzNHFlG9X5TwvvtzvTUPO0pW6bnxaZqBuI+chYP2CP33K4 r2K8VK2SqnIhqDDCPw/wAPNWJyN5JraKxFSlo0ghHX5VJlpO8Fw8Zc+F4e7eRC6FCkFl rvyKHOQqcmeSvXe7aYNAfoHk6SYu+iA8Z6hN68F/Kk4DaRcbpvH2mmZq5IMxSA4WGcSQ qrAA== MIME-Version: 1.0 X-Received: by 10.180.95.162 with SMTP id dl2mr10144888wib.17.1387593205787; Fri, 20 Dec 2013 18:33:25 -0800 (PST) Received: by 10.217.123.69 with HTTP; Fri, 20 Dec 2013 18:33:25 -0800 (PST) Date: Fri, 20 Dec 2013 18:33:25 -0800 Message-ID: From: Dave Taht To: "cerowrt-devel@lists.bufferbloat.net" Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Subject: [Cerowrt-devel] increasing the *codel target for DSL at really low bandwidths X-BeenThere: cerowrt-devel@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: Development issues regarding the cerowrt test router project List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 21 Dec 2013 02:33:28 -0000 I have struggled with really low bandwidths. Folk like fred have really struggled with low bandwidths (to the point of switching to pie on his workload, which has a 20ms target), and having got some configuration info from maxime over free.fr (biggest ecn enabled fq_codel'd deployment I know of), and finally having reasonable confidence that the htb to atm cell size compensation stuff is *fixed*... I'd like to suggest to those with bandwidths under 5mbit try something like this to calculate a larger target variable for *codel. (might make sense on pie, too) There are many significant variables at these speeds, not least of which are fixed constants like iw10. It would not surprise me if we needed to ramp up the curve even further below 1mbit, or double it from what stands here. This curve is good down to almost 128kbit before it overshoots the interval, which regardless needs to be modified too... suggestions? Who here runs at under a mbit on any form of technology? #!/bin/sh # 13000000us for 1 1500 byte packet at 1kbit. bw_up=3D$1 delay=3D$((13500000/($bw_up))) [ "$delay" -lt 5000 ] && delay=3D5000 echo target ${delay}us --=20 Dave T=E4ht Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscribe.= html