Date: Sat, 3 Mar 2007 05:40:35 -0800 From: Luigi Rizzo <rizzo@icir.org> To: John Mok <jmok@attglobal.net> Cc: freebsd-ipfw@freebsd.org Subject: Re: Dummynet cascade of pipes Message-ID: <20070303054035.A78010@xorpc.icir.org> In-Reply-To: <45E96725.3050007@attglobal.net>; from jmok@attglobal.net on Sat, Mar 03, 2007 at 08:16:37PM %2B0800 References: <45E8D473.5090808@attglobal.net> <20070303025618.B76051@xorpc.icir.org> <45E96725.3050007@attglobal.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Mar 03, 2007 at 08:16:37PM +0800, John Mok wrote: ... > Without hierarchical control, would it be possible to make a dummynet > model for the example situation to work? If separate pipes are used to > set the bandwidth limit :- > > ipfw pipe 110 config bw 16 Kbps > ipfw pipe 120 config bw 256 Kbps > ipfw pipe 130 config bw 2048 Kbps > ipfw queue 130 config weight 384 pipe 130 .... > ipfw queue 140 config weight 224 pipe 130 .... > > and when the traffic is fully loaded, then the traffic 2320 Kbps (16 + > 256+ 2048) will exceed the available bandwidth 2 Mbps and would result > in uncontrollable packet drop due to queue overflow. i am not sure about what you want to achieve (because pipes 110 and 120 are unused in your example). Anyways with WFQ when the total exceeds the available bandwidth, queueing will only occur for those flows exceeding their share, others will keep going with their reserved share. If e.g. you have three flows and want to give them at least 16, 256, 1024 kbps (respectively), you can set a pipe with (16+256+1024)=1296kbps and set three queues with weight 1, 16 and 64 (the weight can be in the range 1..100 i think). cheers luigi
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070303054035.A78010>