[142405] in North American Network Operators' Group
Re: IPv6 words
daemon@ATHENA.MIT.EDU (Joel Jaeggli)
Fri Jun 24 11:16:57 2011
From: Joel Jaeggli <joelja@bogus.com>
In-Reply-To: <20110624135024.GB85069@ussenterprise.ufp.org>
Date: Fri, 24 Jun 2011 08:16:38 -0700
To: Leo Bicknell <bicknell@ufp.org>
Cc: NANOG list <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
On Jun 24, 2011, at 6:50 AM, Leo Bicknell wrote:
> In a message written on Fri, Jun 24, 2011 at 09:10:53AM +0000, Bjoern =
A. Zeeb wrote:
>> If you want to do it, make sure you do understand the restrictions =
that apply to IPv6 addresses, like U/G bits, etc. Too many people =
unfortunately just think it's cool in a weird geeky sense and violate =
RFCs with them. I was very close to write an article about that after =
W6D...
>=20
> Perhaps I missed something in an RFC somewhere, but I believe those
> bits only have meaning locally on an Ethernet LAN. They have no
> meaning when used on non-Ethernet networks, for instance POS or on
> a Loopback. If someone wanted to use them for a /128 virtual for
> their web site for instance that would be ok.
>=20
> Or, turning that around, if you assume an IPv6 address is part of a =
/64
> on an Ethernet network, you have made a false assumption.
A load-balancer attached to it's first hop router via a /126 may well =
advertise the virtual ip's it's serving (and treat them) as /128s. the =
assumption that links are /64s falls down a lot (even on ethernet) when =
most of them are point-to-point.
> --=20
> Leo Bicknell - bicknell@ufp.org - CCIE 3440
> PGP keys at http://www.ufp.org/~bicknell/
