Date: Tue, 03 Nov 2009 11:07:10 +0700 From: Eugene Grosbein <egrosbein@rdtc.ru> To: Barney Cordoba <barney_cordoba@yahoo.com> Cc: freebsd-net@freebsd.org, rihad <rihad@mail.ru> Subject: Re: dummynet dropping too many packets Message-ID: <4AEFAC6E.5080805@rdtc.ru> In-Reply-To: <274298.85978.qm@web63908.mail.re1.yahoo.com> References: <274298.85978.qm@web63908.mail.re1.yahoo.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Barney Cordoba wrote: > Your problem is that at high traffic levels you need to reduce traffic > flows, not just delay it as dummynet does. Dummynet does not "just adds delay". > The entire point of traffic > shaping is to smooth out your traffic flows; not to make it so choppy > that you have packets sitting in a transmit queue for 1/2 millisecond > in addition to the dummynet delays. While dummynet may not be dropping > packets, you have packets being dropped in TCP stacks throughout your > customer base, most likely. If you followed the thread, you known that rihad tried GRED. The problem was not due to exceeded bandwidth but in inadequate interface-level FIFO queue length. And no way to adjust it without a patch. This makes me think we should have general user interface for setting the queue length for any network interface just like Cisco 'hold-queue' command does. For now, only some drivers (e.g., em(4)) have such option.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4AEFAC6E.5080805>