[ppml] The myth of IPv6-IPv4 interoperation, was: Re: Legacy /24s

William Herrin arin-contact at dirtside.com
Mon Sep 3 11:15:45 EDT 2007

On 9/3/07, Iljitsch van Beijnum <iljitsch at muada.com> wrote:
> It looks like you suffer from the (fairly common) misconception that
> goes along these lines:
>      "If only the IETF had made IPv6 interoperable with IPv4,
>      we wouldn't have any transition issues."
> The assumption here of course is that it would have been POSSIBLE to
> build IPv6 in such a way that an IPv6 host can talk to an IPv4 host.


That's nonsense. Perhaps you misunderstand the character of backwards
compatibility. Its not a deep technical concept. It has much more to
do with technical and administrative policy decisions and it could
have been implemented with the IPv6 protocol that we actually created.

Consider the following theoretical ruleset:

1. IPv4 addresses map to the IPv6 address space by placing zeros at the front.

2. Any router which receives an IPv4 packet for a destination that is
only served through an IPv6-only trunk must translate the IPv4 packet
to a corresponding IPv6 packet.

3. Any router which receives an IPv6 packet for a destination that is
only served through an IPv4-only trunk must translate the IPv6 packet
to a corresponding IPv4 packet.

4. Any host which implements both IPv4 and IPv6 must consider a return
packet of the other type than the originated packet to be a valid part
of the same communication and should attempt to synchronize with the
same IP version as the remote host.

What are the compatibility consequences of such a ruleset?

A. Once I upgrade the IOS on my Cisco and install the patch from
Microsoft, all of my equipment automatically operates with the new
protocol. There are no new IP addresses to obtain and no configuration
changes to make.

B. The old sockets API works until you want to initiate a connection
to an address that's beyond the scope of the IPv4 addresses. It can
even originate IPv6 packets and accept connections from high-order
addresses, albeit with faulty logging.

C. You can't accomplish most of the subsidiary goals that IPv6 was
intended to accomplish. But then, none of them actually panned out
opertionally anyway.

That's a pretty high level of compatibility. And that's just with the
IPv6 protocol we have. Even greater compatibility gains are possible
with something like http://bill.herrin.us/network/ipxl.html

Bill Herrin

William D. Herrin                  herrin at dirtside.com  bill at herrin.us
3005 Crane Dr.                        Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004

More information about the ARIN-PPML mailing list