IPv6 Pain Experiment

George Michaelson ggm at algebras.org
Thu Oct 3 01:30:30 UTC 2019


A long time ago, in another country, JANET had a mail list to discuss
email, in a world before DNS. And, when DNS emerged, JANET mail list
made a *deliberate* decision to make the domain order of UK email
domains the reverse of every other country worldwide. A DELIBERATE
decision. (I was there, on this list. Others may disagree with my
interpretation of acts done and motivations, but I want to be clear I
didn't "hear this second hand" -I was receiving the mailflows
discussing this in public. I am sure there are other private
conversations I didnt see)

It wasn't a consensus decision. It wasn't an entirely rational
decision. OTOH it was a research network, email was a research
activity, and in some ways, it made sense to find out what happened.
That the decision had repercussions which echoed down the years, and
marginalized some communications Uk and internationally, is perhaps,
the real lesson.

IPv6 had an opportunity to consider designs which were intermediate,
(IPv4-and-a-bit) and backwards compatible. And, like JANET and domain
order, people decided not to do it, believing it was interesting and
research-y.

I too wish we had selected TUBA, or had thought more about interop
with IPv4. I sometimes wish I understood why SRC was the first element
off the wire, and not DST, Since rational ASIC/FPGA hardware can latch
early on the SRC and begin routing faster if it appears in natural bit
order first. Or, why we even have SRC in the header: it does not
inform routing. These are heresies.

Counterfactuals dog Historians. Some love them, some hate them. We
don't have time machines. This is the world we live in, we have to
make the best of it we can. IPv6 globally is rising, IPv6 in Asia is
rising. IPv6 in India is basically ubiquitous, IPv6 in America is
ubiquitous. We are going to live in a mixed protocol global internet
for the forseeable future. We can plan to extend V4, or end V4, or
deprecate V4, or end v6, and favour CGN  but we can't end either V4 or
V6 entirely, easily, soon.

-G



More information about the NANOG mailing list