Raul Echeberria wrote:
At 08:32 p.m. 29/07/2007, David Conrad wrote:
Raul,
On Jul 26, 2007, at 10:04 AM, Raul Echeberria wrote:
In fact I don't think that any RIR promote the use of NAT.
Historically, NAT use hasn't needed promotion.
This is a byzantine discussion and I guess that
we agree regarding the use of NAT.
Paradoxically, the use of NAT in IPv4 is about to become very critically
important to IPv6, I'd suggest.
Most of the "transition" plans we have seen so far for this transition
to an IPv6 network pass through this intermediate stage of dual stack
deployment. The basic idea is that what is required for a V4 host
initiate, maintain and close a "conversation" with a V6 host and
vice-versa goes well beyond the conventional mode of packet protocol
header substitution, and efforts to perform various permutations of
protocol header translators, DNS manipulations and application level
gateways all appear to have their dark and ugly side in terms of cost,
deployment complexity, service fragility and fractured application
transparency.
So, with help from various forms of tunnelling support to bridge over
any protocol-specific transport continuity gaps, the basic idea of this
transition is that we enter an extended period where hosts need to use
V4 to talk to V4 hosts and V6 to talk to V6 hosts. As long as hosts
first try the V6 handshake, then, so goes the line of reasoning, we
should see the traffic mix in this dual stack tend to move to v6 as the
legacy V4 infrastructure migrates to V6, and the dual stack nature of
the deployment means that dual stack hosts can fall back to V4 to speak
to V4 legacy infrastructure.
Nice plan, as far as it went.
Of course the assumption behind this transition was that we'd get moving
with this V6 transition while there was still ample V4 addresses to fuel
the continued growth of this dual stack world, or, minimally, to fuel a
NAT'ted version of the V4 world, and that we'd be done before anything
seriously hit the wall of exhaustion. Some years back this probably
seemed like a sensible assumption.
But the real question we face now is: "How do we support the growth of
the V4 part of this dual stack Internet with a looming exhaustion of the
V4 unallocated address pool?"
V4 NATs may well be very important components of this forthcoming dual
stack transitional world, like it or not.
regards,
Geoff