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

[Mark Smith <[e-mail]>]

On Sat, 26 Oct 2019, 12:30 Brian E Carpenter, <[e-mail]>
wrote:

> >>     Renumbering is not "common" but it certainly happens.
> ...> People imply it is a regular and routine event...
>
> Sadly, it is. For example, at my previous apartment there was a badly
> waterproofed joint somewhere between me and the nearest street cabinet, so
> for months I frequently lost ADSL synch... and when it came back, I got new
> IPv6 prefix. After a heavy rainstorm, this would happen many times a day.
>
> Yes, OK, there was a copper cable fault and it took <unmentionable
> company> many months to fix it. OK, maybe the ISP shouldn't regenerate a
> prefix every time. But that's real life and I suspect I am one of millions.
>

This smells of the local BRAS pool dial up model.  In the same situation,
that wouldn't happen to me because I have a stable /56 associated with my
user account.

If you're using PPPoE, you can specify the access concentrator you want to
(re)attach to.

One common AC/BRAS/NAS implementation caches IPv4 pool assignments to user
accounts, so that if the same account re-attaches, the same IPv4 pool
address is given back to it. I wouldn't be surprised if that also occurs
with IPv6 pool delegated prefix assignments.





> Regards
>    Brian
>