Re: [v6ops] Report: Bar BoF on a 6to4 replacement

[Jeroen Massar <[e-mail]>]

On 2014-12-03 19:14, Keith Moore wrote:
> I followed up to Jeroen in detail in private mail.

Why not keep it on list?

> Executive summary:  He seems to think there's no need for an
> easy-to-configure way to enable v6 connectivity for arbitrary hosts or
> consumer networks because those people who can't be bothered to install
> and configure SixXS just want to look at pictures of cats anyway.

Clearly that is what you interpreting from what I am trying to get out
of you. But that is not what I have stated at all and is actually gross
misrepresentation.

First of all, I do not limit to SixXS (which is just an implementation
to solve a particilar set of problem with a specific demographic
audience), let alone to Tunnel Brokers for finding a solution for the
problem.


One of the major problems with your mails is that the demographic that
you want is completely unclear. As you say "most people", the only
answer can be "folks using HTTP" (thus watching cats, facebook, gmail
and the like). These folks do not even care about SMTP as their email is
a website and they do not know better.

Those people really do not care about IPv4/IPv6 etc as for them "The
Internet" (as they know it) just works fine, even if they are behind
fifty layers of NATs or heck are getting native IPv6-only already.


Hence, please define your demographic and then define the problem they
are having and where they are having that problem.

As only then will we be able to determine where the problems are and
then go for a solution to solve that problem.


> I disagree, and think that large numbers of ordinary users benefit from
> availability of applications other than the web, which run better over
> v6 than with ad hoc mechanisms for NAT traversal.

Like you I love and want that people can use the FULL Internet even
though most people will never know the full possibilities that it offers.

Hence, why I have been deploying IPv6 for so long already: to get people
to be able to deploy things. (of course I wanted to do that myself too,
just happened to end up doing it for others too ;)

But the people who specifically need these kind of setups either read
magazines like the paper editions of www.heise.de/ct/ and
www.heise.de/ix/ or PC Magazin etc who have been covering IPv6 for years
already; or those people google/yahoo/bing and find a solution to their
problem, typically not even a IPv6 tunnel or other form of connectivity
but things like getting a VPS, TeamViewer, Hamachi or a form of a
peer-to-peer network (and then I mean bittorrent not true p2p which is
what one gets with unfirewalled IPv6).

Any other demographic simply does not care about IP.

If you have examples of situations of people who do care about having
proper connectivity to the Internet, then please name them and let us
define them and then see what kind of problems these people have.



> I also think that the
> widespread support of anycast 6to4 in hosts and consumer routers was
> evidence of at least a perception of a need for easy v6 connectivity
> over v4.

Yes, there was this. But we learned that Anycast 6to4 causes problems
that we cannot solve as there are too many layers of operation involved.

Also, since 6to4 came out there have been better solutions that solve
that problem much better. Remember that when 6to4 was developed the only
NAT that existed was the one in the computer of the user, not in the
network of the ISP. There where no mobile networks either and most folks
had dial-up.



> Since anycast 6to4 has been demonstrated to not work well and
> 6to4 with an explicitly configured router still has other problems, I
> want to define a different mechanism that works better but still has
> 6to4's ease of configuration.

But why define something if you cannot even define who will be using it
let alone the problems they currently have?

I'll ask again:
 Which exact gap are you trying to solve that existing methods do not
 solve? And of course, please do specify the demographic.


> I am however happy if all we need for
> that mechanism is a way that networks can advertise, and hosts and
> consumer routers can discover, available mechanisms for tunneling v6
> over v4, and if the hosts/routers can be configured to use a fallback
> tunnel service if no other means is available.

6rd uses DHCP for 'advertising', this works great on the local media to
which an ISP has access.

Anything else will require somebody to run a service where possibly
millions of people can get coordinated access to that service.
That system will not happen as that has the same problem as 6to4
anycast: other entities are paying for that service.



Note that the 'advertising' of tunnel brokers happens in printed media /
blogs and maybe strangely more importantly in the devices that people buy.

Signup reasons that I see regularly when going through the SixXS signup
queue prove this:
 "My Sophos UTM pointed out to this"
 "UnityMedia helpdesk sent me here"
 "My ZyXEL supports SixXS"
 "I saw AICCU support in OpenWRT"
 "I installed the aiccu package"
etc

(Oh, fun side-note: SixXS is just a hobby project that takes too much
time.... hence no real advertising in any way)


> The advertising
> mechanism should be able to advertise 6rd, AYIYA, and anything else that
> works well.

Any such "advertising" mechanism would require central coordination,
such a mechanism does not exist on the Internet and cannot be
distributed easily around the Internet in a scalable method.

Somebody has to run it, there is no way to make something run unattended
and unmanaged and without it costing anybody anything.

Greets,
 Jeroen