RE: (v6ops) WG Last Call: draft-ietf-v6ops-renumbering-procedure-00.txt (fwd)

["Christian Huitema" <huitema@windows.microsoft.com>]

 > Having two prefixes at the same time lands you squarely inside the
> domain of draft-huitema-... 

I actually discussed this point when reviewing an early version of the
draft. Fred took the comments into account in section 2.3, "Configuring
network elements with the new prefix", and in section 3.3, "Ingress
filtering". The text is OK, although it would perhaps be desirable to be
a bit more explicit in section 3.3, since there is a way to shoot
yourself in the foot. The draft's assumption should be made very clear:

* you can shoot yourself in the foot if you route packet to an ISP and
that ISP drops them because they fail the ISP ingress filtering test;
* we must definitely assume that the new ISP will accept traffic sourced
from the old prefix;
* the transition works much better if the old ISP accepts traffic from
the new prefix, in which case there is no risk of any holes in anyone's
foot;
* if the old ISP does not accept traffic from the new prefix, then the
routing should be updated first: send all the traffic to the new ISP
before starting to configure the new addresses.

The case that Fred studies is less constrained than the general multi6
problem. Since the goal is to move from a state where all traffic is
sent to the old ISP to a state where all traffic is sent to the new one,
there is little need to emphasize traffic engineering or load balancing
between the two ISP.

-- Christian Huitema