Re: [IPv6] Adoption call for draft-bctb-6man-rfc6296-bis

[Naoki Matsuhira <[e-mail]>]

Thank you for sharing the difficulty of obtaining PI IPv6.

I understood that there is a need for a method to implement both
multihoming using PI and multihoming without PI.

Naoki

2024年3月26日(火) 20:06 Michael Richardson <[e-mail]>:

>
> Naoki Matsuhira <[e-mail]> wrote:
>     > I understand that there have been many discussions over the years,
> and I
>     > think this is a very difficult issue. In the IPv4 era, there was no
> choice
>     > with assuming exhaustion of IPv4 addresses, but in the IPv6 era,
> there are
>     > plenty of addresses. If so, I think we should make use of our NAT
>     > experience from the IPv6 era.
>
> The rfc6296-bis document has short introduction that goes into the why.
> As you say, it's not about address exhaustion.
>
> I find that the applicability section (1.2) says a few small things and
> then
> gets into a feature/benefit analysis rather than a "when" to use this.
> I think that section 1.1 does a better job at what it sets out to do.
>
> I've said for over ten years that it's just way too hard to get PI IPv6.
> Particularly in the ARIN region.   I would be much happier with this
> document
> and technology if SME's could easily get PI IPv6.
> I don't mind if they have to NPTv6 it when their primary internet
> connection
> dies. (It might be annoying to applications though)
> It could be that many will arrange with all their ISPs to announce it, and
> some will worry that we'll blow up the DFZ.  Maybe.
>
> There have been other proposals, for instance the geographically assign IP
> address scheme from Tony Hain:
> https://datatracker.ietf.org/doc/draft-hain-ipv6-geo-addr/
> There are a number of ways to use it.  Probably never appropriate for
> single
> occupant homes, because of privacy.  But 100 storey buildings might be okay
> with it, and certainly SMEs will not have the same concerns.
> But a brilliant part about it is that far from that location, the addresses
> naturally aggregate.   At the time, we imagined that address to be the
> primary one for connection establishment, and then SHIM6 could negotiate
> some
> number of PA addresses.  We can't do that with SHIM6 (yet?!), but we could
> do
> that with QUIC and/or MPTCP.
>
> If this document is adopted and published, would you support adding a
> signaling mechanism to indicate that NPTv6 is in use?
>
>
> --
> Michael Richardson <[e-mail]>, Sandelman Software Works
>  -= IPv6 IoT consulting =-                      *I*LIKE*TRAINS*
>
>
>
>