From owner-freebsd-ipfw@FreeBSD.ORG Sat Feb 12 05:00:18 2011 Return-Path: Delivered-To: freebsd-ipfw@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id E6A64106564A for ; Sat, 12 Feb 2011 05:00:17 +0000 (UTC) (envelope-from nuno@diogonet.com) Received: from mail-fx0-f54.google.com (mail-fx0-f54.google.com [209.85.161.54]) by mx1.freebsd.org (Postfix) with ESMTP id 7CAEB8FC0A for ; Sat, 12 Feb 2011 05:00:16 +0000 (UTC) Received: by fxm16 with SMTP id 16so3708722fxm.13 for ; Fri, 11 Feb 2011 21:00:16 -0800 (PST) MIME-Version: 1.0 Received: by 10.223.86.13 with SMTP id q13mr6659029fal.53.1297383832619; Thu, 10 Feb 2011 16:23:52 -0800 (PST) Received: by 10.223.107.19 with HTTP; Thu, 10 Feb 2011 16:23:52 -0800 (PST) In-Reply-To: References: <20110210072811.GA41597@onelab2.iet.unipi.it> Date: Thu, 10 Feb 2011 19:23:52 -0500 Message-ID: From: Nuno Diogo To: nangergong Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Cc: freebsd-ipfw@freebsd.org, Luigi Rizzo Subject: Re: about "profile" in IPFW/dummynet X-BeenThere: freebsd-ipfw@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: IPFW Technical Discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 12 Feb 2011 05:00:18 -0000 Just wanted to highlight Luigis point that the profile method of adding delay, refers to MAC overhead delay not latency/RTT. I asked him about this a while ago when I was trying to do doing what you are doing, and what you'll find is that application performance with 200-1300ms profile delay will be MUCH worse than with 200-1300 propagation delay. This is because with profile, the delay is applied before the packet gets queued in the pipe, making the hop unavailable for that amount of time, while the 'normal' delay setting is applied after the packet is queued in the pipe, simulating delays that occur after the packet leaves the bottleneck. Unfortunately I still haven't found a good way to emulate variable RTT with dummynet. On Thu, Feb 10, 2011 at 9:35 AM, nangergong wrote: > thank you! > I add "bw" in the distribution.txt. However, the problem remains. > when I ping 9.161.148.72 from 9.161.150.55, the RTT<1ms > are there any other problems? > > > ----------------------------------------------------------- > name =A0 =A0distribution > bw 2Mbit/s > samples 100 > loss-level 0.86 > prob =A0 =A0delay > 0 =A0 =A0 =A0 200 =A0 =A0 # minimum overhead is 200ms > 0.5 =A0 =A0 200 > 0.5 =A0 =A0 300 > 0.8 =A0 =A0 1000 > 0.9 =A0 =A0 1300 > 1 =A0 =A0 =A0 1300 > #configuration file end > > > > > On Thu, Feb 10, 2011 at 7:28 AM, Luigi Rizzo wrote: > >> On Thu, Feb 10, 2011 at 12:09:09AM +0000, nangergong wrote: >> > Hi, all: >> > >> > =A0 =A0I want to use "profile" to simulate delays according to a empir= ical >> delay >> > distribution ( the "profile" argument can be found in >> > http://fuse4bsd.creo.hu/localcgi/man-cgi.cgi?ipfw+8) >> > I use the following command lines and distribution.txt, and find that >> these >> > command lines seem not to =A0function. when I ping 9.161.148.72, the >> RTT<1ms >> > >> > My question is >> > 1) can "profile" be used in windows platform >> > 2) are there any problems in my procedures? >> >> you need a non-zero bandwidth if you want to use "profile". >> Also, "profile" emulates an additional transmission time >> (think of channel arbitration and mac overheads) >> not propagation delay. >> >> Finally I believe profile works on all supported platforms. >> In general, the differences between FreeBSD/linux/windows are >> only in areas where certain kernel features are implemented >> in different ways on different platforms -- e.g. sysctl, >> reinjection, matching on pid/gid or other credentials, divert. >> >> cheers >> luigi >> > >> > >> > my command lines are: >> > >> > ipfw pipe 1 config profile distribution.txt >> > ipfw add 100 pipe 1 ip from 9.161.150.55 to 9.161.148.72 >> > >> ------------------------------------------------------------------------= -------------- >> > if I input: ipfw pipe show >> > >> > the result is: >> > >> > 00001: unlimited =A0 =A0 =A0 =A0 0 ms burst 0 >> > =A0 =A0 =A0 =A0 =A0profile: name "distribution" loss 0.850000 samples = 100 >> > q131073 =A050 sl. 0 flows (1 buckets) sched 65537 weight 0 lmax 0 pri = 0 >> > droptail >> > =A0sched 65537 type FIFO flags 0x0 0 buckets 0 active >> > >> > >> > the content of distribution.txt is: >> > >> > name =A0 =A0distribution >> > samples 100 >> > loss-level =A0 =A00.86 >> > prob =A0 =A0delay >> > 0 =A0 =A0 =A0 200 =A0 =A0 # minimum overhead is 200ms >> > 0.5 =A0 =A0 200 >> > 0.5 =A0 =A0 300 >> > 0.8 =A0 =A0 1000 >> > 0.9 =A0 =A0 1300 >> > 1 =A0 =A0 =A0 1300 >> > #configuration file end >> > _______________________________________________ >> > freebsd-ipfw@freebsd.org mailing list >> > http://lists.freebsd.org/mailman/listinfo/freebsd-ipfw >> > To unsubscribe, send any mail to "freebsd-ipfw-unsubscribe@freebsd.org= " >> > _______________________________________________ > freebsd-ipfw@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-ipfw > To unsubscribe, send any mail to "freebsd-ipfw-unsubscribe@freebsd.org" > --=20 ---------------------------------------------------------------------------= ---------------------- Nuno Diogo