ietf
[Top] [All Lists]

Re: IPv6 addresses really are scarce after all

2007-08-28 11:11:53

On  21 Aug 2007, at 22:54, Jun-ichiro itojun Hagino wrote:
IEEE 802 standards do not permit variation in the link MTU
for Ethernet.  Attempts to persuade IEEE 802 to approve use
of jumbo-MTUs (e.g. 9180 bytes + Ethernet framing) have
consistently failed within the IEEE 802.

        ok, then pls think about FDDI-to-ethernet bridge
        (i guess it is also spec conformant but there are products),
        and/or 802.11 bridges.

My understanding is that 802.11 has no difference
in Link MTU size from 802.3.  If someone thinks that
is incorrect, citing the specific IEEE standards
and the specific section numbers from those standards
would be helpful.

FDDI chipsets ceased manufacture around 2002, so
FDDI has not been available for purchase for several
years already.   Almost all the FDDI in the world
has already been removed by now.  (US Navy had been
a very large FDDI user and they started work on removing
all of their FDDI around 2000/2001.)

I think no one has FDDI in their house LAN or small-office
LAN anyway; house LAN and SOHO LANwas the focus of my posting.
Large company networks will not have difficulty obtaining
a prefix shorter than /64 from a commercial ISP that supports
IPv6, so those are not an issue either.

Token Ring, FDDI, and Token Bus are almost non-existent
today in the deployed world -- and never had significant
deployment in home LANs or small-office LANs.

So it seems the only potential issue with link MTU compatibility
in the real world is from Jumbo Ethernet (9180 bytes + Ethernet
framing overhead) -- and Jumbo Ethernet is explicitly violating
the IEEE 802 standards.  (Some very-lost-cost Ethernet switch/hub
or router products can support Jumbo Ethernet, but I believe
that feature defaults to OFF.)

Cheers,

Ran
rja(_at_)extremenetworks(_dot_)com




_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf