[30252] in North American Network Operators' Group
Re: exchange point media
daemon@ATHENA.MIT.EDU (Simon Leinen)
Fri Jul 21 14:33:45 2000
To: Mikael Abrahamsson <swmike@swm.pp.se>
Cc: nanog@nanog.org
From: Simon Leinen <simon@limmat.switch.ch>
In-Reply-To: Mikael Abrahamsson's message of "Mon, 17 Jul 2000 23:32:20 +0200 (MEST)"
Date: 21 Jul 2000 20:30:16 +0200
Message-ID: <aa1z0nwd3b.fsf@limmat.switch.ch>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Errors-To: owner-nanog-outgoing@merit.edu
>>>>> "ma" == Mikael Abrahamsson <swmike@swm.pp.se> writes:
> I was under the impression that there is nothing magical about jumbo
> frames and that there are no interoperational problems with them as long
> as they're supported at all. Please correct me if I am wrong.
I'm afraid you are wrong... all interfaces connected to a logical IP
subnetwork should have the same MTU. For most Layers-2 there is a
pretty well-established default such as 1500 for Ethernet, 4470 for
FDDI (well for ATM many people use 4470 rather than the IETF
suggestion of 9180). But for Gigabit Ethernet the only standard MTU
is the Ethernet MTU (1500).
For a new exchange point built on Gigabit Ethernet, it could be
worthwhile to postulate a larger MTU such as 8192 or 9180 and make
sure every participant actually uses exactly that.
But most exchange points supporting GigE still provide bridging to
legacy (Fast) Ethernet, on which larger MTUs are uncommon or not
supported at all. And bridging between Layer-2s is similarly
problematic to having different MTUs in a LIS, although I'm well aware
that such configurations are being used in many places, and thanks to
bridges that do clever stuff besides bridging this may even work more
or less reliably.
--
Simon.
