An Internet IPv6 Transition Plan

Chad Oleary oleary.chad at gmail.com
Tue Jul 24 14:01:44 UTC 2007


I'm wondering if we should really be considering a "transition" plan
at this point? From what I can see, there will be many IPv4 only
networks around for many years to come. The technology doesn't have an
expiration date. Rather than focusing on "transitioning" every network
from v4 to v6, shouldn't this draft focus on how an "initial
implementation of v6" is going to interop with v4 as we know it today?

Personally, I see v6 as something that needed and desired by the
certain groups. However, when looking at the enterprise, for example,
better solutions are needed for things like multi-homing, last I
checked. IPv4 will get more expensive as time goes on, but some will
be willing to pay that price.

Perhaps the biggest challenge, IMO, in this much more dynamic network,
is DNS. How do I (or my new vendor) readdress every node at my site,
and actually know what device has what address? rtadvd doesn't do DNS
updates. DHCPv6 doesn't even hand out addresses. I've seen host-based
approaches, is that the answer? How does all this happen securely?
DNSSEC comes to mind, but that's a whole different story. Add, since a
host can have many preferred addresses, which to use? How do
deprecated addresses get withdrawn from DNS?

I think a more successful approach would be to address how we plan to
add v6 to the current network. Perhaps a transition plan is
appropriate for some networks. But, I don't think this is a one-size
fits all issue.

This is the part that I have issue with:

 2.3.4 Service Providers area MAY continue to offer IPv4-based Internet
       connectivity to their customers.  Organizations MAY continue to
       use IPv4-based Internet connectivity.  Organizations MAY remove
       IPv4-based Internet connectivity from Internet-facing servers.

If I'm an IPv4 only site outside of this "perfect world", I just lost
connectivity to parts those that moved to IPv6. Not everyone will
follow this plan, and this will happen. For now, we need to learn how
to co-exist.

Thanks for your time,

Chad



More information about the NANOG mailing list