How far has the Internet come with IPv6 Adoption?

[Martin Millnert]

On Wed, 2011-04-20 at 08:25 -0700, Ted Mittelstaedt wrote:
> On 4/20/2011 7:50 AM, JORDI PALET MARTINEZ wrote:
> > I've not read the study, but it is normal that there is no native IPv6
> > traffic,
> 
> they didn't say there was NO native IPv6 traffic.
> 
> and it will not be there for long time, possibly until we reach
> > 40-50% of ISP, deploying dual stack in the last mile.
> >
> > If they measure transition traffic, the picture will be much different,
> > because the peer-to-peer traffic.
> >
> 
> They did.  If you think there's a problem with the study then
> have ATLAS send you a darknet probe and stick it on your network.
> 

If the problem is with the methodology itself, sticking a darknet probe
on your network is not going to help.

I'd like to know more about the methodology used to identify Teredo
traffic, because I'm still not convinced their Teredo-detection
algorithm is correct.  "Microsoft’s Teredo (65.55.158.118)." is not very
convincing to this end.  There are certainly more than 1 IP used by the
different MS Teredo server farms -- and MS does not run any relays at
all by themselves...
  By mentioning one specific MS IPv4 address as "Teredo", they fail to
suggest that they've done anything close to
http://tools.ietf.org/html/draft-ietf-v6ops-teredo-security-concerns-02#section-3.2.2 which is required to properly identify Teredo data payload.

>From operating a 6to4- and Teredo relay once, I found that the traffic
levels of Teredo were outperforming 6to4, once the Miredo setup was
optimized a bit, and that there was no upper limit in traffic levels of
Miredo -- the more I optimized, the more throughput grew, suggesting the
relay was a bottleneck in the data path even though no pl was measured.

I can only assume from the report that they've (again) completely missed
~99.999% of the true Teredo volume, if they indeed only looked at
traffic to that one IP.  There is no independent presentation of 6to4 vs
Teredo traffic to draw any further conclusions from wrt. how
successfully they've classified Teredo data payloads.

Regards,
Martin