Re: A common problem with SLAAC in "renumbering" scenarios

[Tom Herbert <[e-mail]>]

On Tue, Feb 19, 2019 at 12:23 PM Manfredi (US), Albert E
<[e-mail]> wrote:
>
> -----Original Message-----
> From: ipv6 <[e-mail]> On Behalf Of Ole Troan
>
> > On 19 Feb 2019, at 14:07, Mikael Abrahamsson <[e-mail]> wrote:
> >>
> >> Indeed. Wonder how these pesky mobile phone operators manage to deliver the same telephone number to a user, for years. Across different providers and contracts.
> >
> > They use a directory to convert between the stable identifier used, and the temporary one that can be changed, so that the stable identifier is fairly easily handled by humans. If only we would have that...
> >
> > Oh, wait... we do. DNS.
>
> Exactly. An extra layer of abstraction is introduced in mobile telephony, and the "telephone number" now becomes merely a unique identifier, not used for routing. So the example is much more similar to say, what the enum wg did in the past, to convert telephone numbers into IP addresses, and really nothing to do with IP addresses as they relate to routing packets, and never mind the security/privacy issues that have been elaborated for so long, with respect to stable IIDs.
>
I believe in mobile networks it the stable identifier of a device is
an IMSI. It would be a very bad idea to create IP addresses that embed
an IMSI, there is not security or privacy in that. It's similar to how
MAC addresses shouldn't be encode in IP addresses. So there is always
at least some decoupling between identity and address. As the demand
for secuirity and privacy increases, the need for decoupling them
increases as well. Taking that to its fullest extent, maximum privacy
and security happens when addresses are completely decoupled from
identity so that a third party cannot make any inference or
correlations about identity, location, or any other PII just based on
observation of IP addresses in packets. That requires very dynamic
addressing.

Tom

> In short, if the IETF were to push for stable IPv6 addresses to customers, especially mobile customers, then the route aggregation problems don't go away, nor do the security and privacy problems. They become hidden underneath this upper layer of abstraction.
>
> The privacy concern I think we have been trying to alleviate is to make that actual IPv6 prefix, the one used for routing packets, not permanently linked to the same household, for all time. Any abstract, DNS-like, IP address, which is mapped to the actual IPv6 address, is not really the issue here, I don't think.
>
> Bert
>
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [e-mail]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------