Re: [v6ops] SLAAC renum: Problem Statement & Operational workarounds

[Philip Homburg <[e-mail]>]

>I'm skeptical that playing with prefix timers and protocol mechanisms will do m
>ore than put lipstick on the pig. Iff we wanted to solve ephemeral addressing /
> connectivity, that's would require something else.

A typical home situation today has almost exclusively outbound connections.
People also accept that if you reboot a home router, a video stream may
be interrupted.

Where general problem of renumbering is hard to solve, the more limited 
home use case is not. For IPv4 all of this is hidden by NAT in the CPE.

After the previous discussion, I updated my IPv6 stack to do detect prefixes
that are no longer announced in RAs. This does solve the problem at hand.

Previously, AVM CPEs (Fritz!Box) would actively kill weird prefixes used
by hosts. That was quite effective, though has some weird side effects.
So a CPE solution is likely to solve this problem as well.

This is problem that can be solved both in the host and in the CPE.

Of course, we can also wait for ISPs to start shipping CPEs with NAT66.

Reality is that ISPs do flash renumbering for IPv4 customers. We can tell them 
that they cannot do that for IPv6, but that will result either in a delay
of IPv6 deployment, or they will find a hack (such as NAT66) to make it 'work'.