[76788] in North American Network Operators' Group
Re: Smallest Transit MTU
daemon@ATHENA.MIT.EDU (Iljitsch van Beijnum)
Wed Dec 29 14:15:22 2004
In-Reply-To: <a06200700bdf8a49ed769@[65.199.121.152]>
Cc: nanog@merit.edu
From: Iljitsch van Beijnum <iljitsch@muada.com>
Date: Wed, 29 Dec 2004 20:14:47 +0100
To: Jerry Pasker <info@n-connect.net>
Errors-To: owner-nanog-outgoing@merit.edu
On 29-dec-04, at 19:30, Jerry Pasker wrote:
> When encountering networks with MTUs smaller than 1500 bytes, path MTU
> discovery breaks
What you mean is: sites that employ non-first hop links with MTUs
smaller than 1500 bytes experience frequent reachability problems.
> block all icmp, including the icmp "fragmentation needed but DF bit
> set" packets.
And yet they send out packets with the DF bit set...
Life is too short to try to communicate with people like this.
> Despite what the RFCs say, the transit internet, in my opinion,
> generally needs to accept and transit packets up to 1500 bytes without
> packet fragmentation.
1. No. If we start to give in to people who drive a firewall without
proper training we may as well shut down the net now.
2. It does. The percentage of places where _transit_ can't handle 1500
bytes disappears in the error margin.
> Is there an RFC that clearly states: "The internet needs to transit
> 1500 byte packets without fragmentation."??
I very much doubt it, but I haven't read all 3500 of them. Feel free to
write one, though, if you aren't bothered by the irony of writing an
RFC to tell people how to handle the situation where people don't
conform to RFCs.
(FYI: RFC 791 says: "Every internet module must be able to forward a
datagram of 68 octets without further fragmentation.")
(FYI2: In IPv6 there is a mandatory 1280 byte minimum maximum packet
size. Not that this helps unless people turn off PMTUD which would
solve the problem on its own in IPv4 too.)
