--On Friday, September 23, 2011 21:20 -0400 Keith Moore
<moore@xxxxxxxxxxxxxxxxxxxx> wrote:
> I already made one Last Call comment, but I neglected to state
> unambiguously whether I supported the proposal.
>
> I do support this proposal.
>
> I think that this question needs to be viewed as a choice
> between two risks:
>
> 1) the risks associated with this proposal
> 2) the risks associated with reuse of RFC 1918 address space
> by ISPs, and/or reuse of public IPv4 address space by ISPs
>...
> It needs to be understood that at this point, there is no path
> that will avoid widespread breakage of much existing
> IPv4-based software, including some software that is widely
> used. Upgrades to that software will be needed in order to
> continue using such software on a widespread basis.
And that helps identify a third risk. How relevant it is
depends on one's perspective and understanding of reality but
that risk is:
3) People will conclude that these various kludges are
actually medium-term solutions and will put resources
into them that would have gone into deploying IPv6
instead.
As soon as you say "folks are going to need to go to the trouble
and expense of developing and deploying replacements for a lot
of installed-base IPv4 software, the resources involved become
significant and there are tradeoffs with other ways in which
those resources could be invested.
Remembering that an ISP who wants to avoid the use of public
IPv4 addresses on its backbone/infrastructure has the option of
simply converting that infrastructure to IPv6, tunneling
public-address IPv4 packets (both its own and those of its
customers) over that IPv6 infrastructure using a tunneling
approach of its choice. Longer-term, that approach makes the ISP
far more IPv6-ready, while "more private/shared IPv4 space" is
just another dead end.
Without commenting on the merits of this particular proposal
relative to other ways to squeeze a few more addresses out of
the IPv4 space, it seems to me that a lot of the presumed value
is ultimately a cost tradeoff. If the squeezing technique buys
extra addresses and time at very low cost, then it is worth
considering (and comparing to other squeezing-out proposals).
If it requires changing enough deployed software and
infrastructure to start approaching IPv6 deployment costs, it
seems to me to be bad strategy and worse economics.
john
>
> Furthermore, even with such upgrades, the reliability of
> IPv4-based applications can generally be expected to decrease
> over time. There is no path to permit IPv4-based
> applications to continue to be used reliably, at Internet
> scale, over the existing Internet infrastructure.
>
> Keith
>
> _______________________________________________
> Ietf mailing list
> Ietf@xxxxxxxx
> https://www.ietf.org/mailman/listinfo/ietf
_______________________________________________
Ietf mailing list
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf