[85813] in North American Network Operators' Group
Re: And Now for Something Completely Different (was Re: IPv6 news)
daemon@ATHENA.MIT.EDU (Fred Baker)
Mon Oct 17 17:59:08 2005
In-Reply-To: <B8E85690-4C25-4DA8-BC04-55F6369DCEA3@tony.li>
Cc: nanog@merit.edu
From: Fred Baker <fred@cisco.com>
Date: Mon, 17 Oct 2005 14:58:37 -0700
To: Tony Li <tony.li@tony.li>, Randy Bush <randy@psg.com>
Errors-To: owner-nanog@merit.edu
On Oct 17, 2005, at 2:24 PM, Tony Li wrote:
> To not even *attempt* to avoid future all-systems changes is
> nothing short of negligent, IMHO.
On Oct 17, 2005, at 2:17 PM, Randy Bush wrote:
> and that is what the other v6 ivory tower crew said a decade ago.
> which is why we have the disaster we have now.
and there I would agree, on both points.
now, the proposal put forward lo these many moons ago to avoid any
possibility of a routing change was, as I recall, Nimrod, and the
Nimrod architecture called for variable length addresses in the
network layer protocol and the use of a flow label (as in "IPv6 flow
label") as a short-form address in some senses akin to a virtual
circuit ID. There has been a lot of work on that in rrg among other
places, but the word from those who would deploy it has been
uniformly "think in terms of an incremental upgrade to BGP" and
"maybe MPLS will work as a virtual circuit ID if we really need one".
As you no doubt recall all too well, the variable length address was
in fact agreed on at one point, but that failed for political
reasons. Something about OSI. The 16 byte length of an IPv6 address
derived from that as well - it didn't allow one to represent an NSAP
in IPv6, which was an objective.
So the routing problem was looked at, and making a fundamental
routing change was rejected by both the operational community and the
routing folks.
No, IPv6 doesn't fix (or even change) the routing of the system, and
that problem will fester until it becomes important enough to change.
But lets not blame that on the "ivory tower folks", at least not
wholly. We were all involved.
