Date: Thu, 09 Aug 2001 04:43:25 +0900 From: Jun-ichiro itojun Hagino <itojun@iijlab.net> To: Jeroen Ruigrok/Asmodai <asmodai@wxs.nl> Cc: Mike Tancsa <mike@sentex.net>, freebsd-net@freebsd.org Subject: Re: gif MTU of 1280 ? Message-ID: <20010808194325.D1B9E7BB@starfruit.itojun.org> In-Reply-To: asmodai's message of Wed, 08 Aug 2001 18:14:54 %2B0200. <20010808181454.Q2937@daemon.ninth-circle.org>
next in thread | previous in thread | raw e-mail | index | archive | help
>Actually I am wondering about it now myself. X.25 is one of the few >link layer protocols left which has a MTU < 1500 (aside from 802.3's >1492). > >Maybe some IPv6 guru is able to shed some light? we picked 1280 because it is the IPv6 minimum link MTU, and would liked to torture-test IPv6 stack with gif (configured as IPv6-over-IPv4 tunnel). for other usage, you may want to SIOCSIFMTU, or whatever. at this moment gif(4) does not keep track of the path MTU discovered for the outer header. this is because, IP protocol (at least both v4/v6) does not cope well with L2 that changes MTU too frequently. (therefore, gif MTU should better be statically defined) as others noted, if you set gif MTU to 1500, then you will see a lot of fragmented packet exchanges (= poor performance). if you use IPv4 as the outer header, MTU <= 1480 avoids fragmentation and more performant. as 1280 <= 1480, we avoid fragmentation with the current default configuration. itojun To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010808194325.D1B9E7BB>