6to4 stats found, are there other places?

[Fred Baker]

On Apr 17, 2011, at 1:38 PM, William F. Maton Sotomayor wrote:

> It's a pity that 6to4 deprecation is being discussed within the IETF at the moment, because I would like to see the issues fixed instead throwing the thing away (whatever that degree may be).  It's worked for us and for the other R&E networks thus far so I'd hate to see the OttIX one go.

There are actually two things under discussion in the IETF. Brian Carpenter wrote a document on how to fix 6to4, and Ole Troan wrote a document suggesting that RFCs 3068 and 3056 be "off by default".

I think it's at least fair to point out what "deprecate" or "move to historic" means in the IETF environment. When the Internet moved from classful addressing to CIDR in the early 1990's, the IESG "deprecated" RIPv1, moving it to "historic". We (I include myself as I was on the IESG at the time) had no expectation that this would mean people would stop using RIP, or that manufacturers would change their products. It meant that RIPv1 didn't address the needs of the growing Internet, and was not something the IETF found useful to continue enhancing. RIPv2 came along a couple of years later, and included several important changes - it moved from broadcast to multicast, it carried a prefix as opposed to a classful address, and it dropped the aggregation algorithms RIPv1 used across classful address boundaries.

In the IETF, we have a fair bit of evidence that 6to4 isn't all its cracked up to be; take a look at Geoff Huston's discussion, in the IETF-80 proceedings. Basically, he's looking at darknet data, and finds 6to4 and Teredo traffic there, which says that at best the protocols send traffic to places its doesn't intend, and at worst they are components of or carriers of attacks of various kinds.

What I understand the two drafts to be saying is
a) if someone is using 6to4, they should be doing so intentionally, not by accident (troan)
b) They should do so in ways that work (carpenter)
c) As IPv6 deploys in native space, transition technologies in general and 6to4 in specific should atrophy.

Commentary from the operator community would be interesting and helpful in the IETF discussion.

http://tools.ietf.org/html/draft-ietf-v6ops-6to4-advisory
  "Advisory Guidelines for 6to4 Deployment", Brian Carpenter, 31-Mar-11

http://tools.ietf.org/html/draft-ietf-v6ops-6to4-to-historic
  "Request to move Connection of IPv6 Domains via IPv4 Clouds (6to4) to
  Historic status", Ole Troan, 5-Apr-11