On Nov 12,
The report as presented at the RIPE meeting indeed mentions the
possibility of undercounting. However, it appears that there is an
undercount of several orders of magnitude. At that point you really
cannot claim that the report provides a perspective on Internet IPv6
traffic as it does. It is quite reasonable to conclude that something
went wrong with the methodology, measurements or analysis.
Nothing is wrong in the methodology and the places where
undercounting likely occurred (namely: flow types supported
by exporting routers, Teredo data channels, etc..) have been
identified. Caveat-aware, I believe the report to be both
very quantitive and qualitative. Furthermore, what we measured
is what the ISPs involve have visibility to, which is a critical
consideration - if you can't see it, and can't measure it, then
you certainly can't qualify it.
If you have any more *quantitative* and qualitative studies
that you can point to I and many others would be quite
interested.
The difference between something that is barely measurable and
something small but measurable like 0.1% is huge. Basically, 0.1% on
the scale of the Internet means that a very large group of people is
using IPv6 today. There is no question that that group pales to the
total number of Internet users but it sure is more than a few people
in IETF experimenting with IPv6.
That's great news, and I look forwarding to seeing more
data from this large group of people...
To be clear, our attempt with this study was to measure
observable IPv6 traffic in production networks across a
large number of production ISP networks. It was not to
discredit IPv6 in any way, quite the contrary.
I look forward to any credible data that you can provide
to support wider adoption, or being made aware of any
unacknowledged issues with our methodology.
-danny
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf