Date: Sat, 23 Mar 1996 22:14:43 -0800 From: David Greenman <davidg@Root.COM> To: Brian Tao <taob@io.org> Cc: Richard Wackerbarth <rkw@dataplex.net>, FREEBSD-HACKERS-L <freebsd-hackers@freebsd.org> Subject: Re: Changing Ethernet frame size to 576 bytes? Message-ID: <199603240614.WAA02784@Root.COM> In-Reply-To: Your message of "Sun, 24 Mar 1996 00:45:48 EST." <Pine.BSF.3.91.960324004213.8944J-100000@cabal.io.org>
next in thread | previous in thread | raw e-mail | index | archive | help
>On 23 Mar 1996, Richard Wackerbarth wrote: >> >> His suggestion is valid. If the MTU is larger than the minimum MTU >> along the path, each packet that you send out will be broken into >> sub-packets and reassembled on the far end. If any subpacket gets >> lost, the whole packet is lost and must be retransmitted. > > Then why doesn't everyone on the Internet use 576-byte packets? I >assume on a fast link, the larger the packet size, the less overhead >there is in sending each one? I've never heard of an FTP server (or >Web server or news server, since they could all apparently benefit >from this) use anything other than the default 1500 MTU. Just because the ethernet interface has an MTU of 1500 doesn't mean that's what the size the packets are going to be. Traditionally, the MSS (maximum segment size) is 512 for all non-local connections (those hosts which are not directly connected to the ethernet). This was changed recently in FreeBSD when Path MTU Discovery was implemented. We now use a mechanism to "discover" what the largest MSS is without fragmenting the packet. This will be whatever the smallest MTU is of all of the interfaces throughout the route minus the TCP/IP header. -DG David Greenman Core-team/Principal Architect, The FreeBSD Project
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199603240614.WAA02784>