just a pessimistic note

[Dave Taht]

On Thu, Feb 23, 2012 at 2:05 PM, Francis Dupont <fdupont at isc.org> wrote:
>> the persistent connectivity problems we had during the call are all
>> going to be made much worse by the short NAT timeouts and restricted
>> port ranges inherent in this technology. Sadly I kind of view
>> development and deployment  of this technology to be necessary lacking
>> a herculean effort to get full ipv6 deployed.
>
> => do you mean you don't like NATs and CGNs?

I think NAT solved a very real problem by eliminating the need for
co-ordinating with more central authorities to arbitrarily expand
internal networks. IPv6, without nat, certainly demonstrates how
difficult it is to do that.

I don't have an opinion yet on CGNs, aside from what I wrote
previously. I'd like to be building a more reliable, robust internet
and it's been going in the wrong direction for a long while. As an
example, I worked on asterisk/sip from 2001 through 2006 and ran
dozens of connections simultaneously all around the globe for hours on
end in testing without the kind of flakyness we see today.

far too many service providers have arbitrarily short nat timeouts and
nat resets already.

I think we are in for a long, difficult transition to ipv6, and to do
interim solutions as 'good as possible' is in part why I'm interested
in this project.


>> An open question for me is if sip or skype can be made to work at all
>> under this for any reasonable period of time.
>
> => skype should not be a problem as it is able to work through NATs,
> SIP is a dead end: it requires a very complex ALG.

Um, skype tries to run over udp where possible, and so far as I know
doesn't do upnp-nat. (could be wrong).

I have no opinion on the future of sip, it is certainly over-complex,
but in some recent testing over pure ipv6
was the lowest latency and best video calls I'd ever had.
> Regards
>
> Francis Dupont <fdupont at isc.org>



-- 
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://www.bufferbloat.net