Re: [v6ops] The need for local-ipv4 socket transition solutions -- NAT64/DNS64 remains insufficient

[James Woodyatt <[e-mail]>]

On Sat, Apr 11, 2015 at 5:05 AM, Xing Li <[e-mail]> wrote:

>  No. I am an academic network service provider and my goal is to provide
> global IPv4/IPv6 Internet connectivity using an IPv6-only backbone and very
> limited public IPv4 addresses.
>

I can understand the goal of providing service on your networks to global
IPv4 and IPv6 destinations, and I can understand how you may be limited by
the scarcity of public IPv4 address space, but I fail to see either A) how
that goal is best achieved by removing IPv4 service from your local
network, or B) how the negative effects of scarcity in public IPv4 space is
mitigated by the deployment of a CLAT in host operating systems.

>From my perspective, deploying a CLAT ubiquitously in host operating
systems will have two deleterious effects: 1) it will prolong the viability
of IPv4 literals in publicly reachable application services, requiring the
maintenance of the PLAT service, and the public IPv4 routing that entails,
over a longer period of transition, possibly indefinitely; and, 2) it will
slow the transition of applications in your local network to using IPv6
instead of IPv4, because the IPv4-only legacy interfaces will still be a
viable, if somewhat impaired, despite your having removed IPv4 routing from
local subnets.

I would think the negatives of this plan would obviously outweigh the
positives. On the plus side: you get to deploy a nice clean IPv6-only
network that nobody knows is IPv6-only. On the minus side: you may never
get to stop supporting IPv4 in the applications and the hosts, because if
you ever do, the users will howl about it, and this you cannot abide. I
would think it better to just decide you can't ever live without IPv4 and
commit yourself to running dual-stack forever.


-- 
james woodyatt <[e-mail]>
Nest Labs, Communications Engineering