RE: An alternative to 6to4 and teredo

[Pekka Savola <pekkas@netcore.fi>]

On Fri, 17 Jan 2003, Jeroen Massar wrote:
> There is at least Tunnel Setup Protocol (TSP*) which does automatic
> configuration and it'salso quite extendable. On debian for instance
> it's "apt-get install freenet6" and you are going. The only 'problem'
> is that the Freenet6 broker system is located in the US thus european
> hosts have some (80ms+) additional latency. A "POP" per ISP would be
> better which is something we are pursuing for SixXS. The TSP protocol
> can handle this fortunatly. A POP per ISP would at least mean that
> clients can get near-native IPv6.
> 
> TSP: http://www.freenet6.net/draft-tsp.shtml
> It is marked "Expires: November 30, 2001" though, what happened
> that it didn't get pushed through?

Folks probably lost hope.
 
> Good example is xs4all who first had a IPng.nl based tunnelbroker
> and modified it for their needs with which they are now providing
> IPv6 to their clients who can't get it natively. The IPv6-in-IPv4
> thus only is carried in their network untill their tunnelbox,
> for xdsl clients this is thus max 3 hops and a latency of about
> only <~3ms added when they would have native connectivity.
> See the "Deployment at XS4all" presentation to be found at:
> http://www.ams-ix.net/aiad/presentations.html for more info.
> 
[...]
> I've partially developped an alternative to this scheme with some
> big differences to what Freenet6 does with their TSP protocol.
> Some other things are in line first though. But it will be a
> system I want to have deployed at least before march for the
> SixXS system which we fortunatly abstracted with an API so that
> things like these can be implemented quite easily.
> Small example is port 42006 on our noc.sixxs.net box which
> basically replaces the website and can be easily accessed by
> any program capable of sending data over a tcp/ip socket.

Above you show one of the main points here: there are dozens of
tunnelbrokers, and each have their own interface (pretty much).  It's just
too much work that way.

(btw, I think tunnel brokering would get off a bit if a client+server code
using some agreed protocol would happen to be publicized under a liberal
license like BSD.)
 
> As I also probably wrote before, people do also sign up for
> things like MSN and hotmail etc, so an IPv6 tunnel via a
> webinterface shouldn't be that hard either.
> The only thing is making them think that having it is useful!
> (Which is one part I have on the 'other things to do first' :)

Hotmail etc. have been around for ages.  Tunnel brokers come and go.  The 
configuration may require some installation in your computer.  Sure, many 
folks still do it, but there's a big difference.

But of course, some kind of "tunnel broker discovery" or catalogues could
be coined up too...

[moved this from above]
> My point is that ISP's should be pushed to have something like
> that or that at least they should deploy a 'close' relay, may
> this be 6to4 or a tunnelbased system. As long as their clients
> can connect as 'locally' as possible.

This is not all.

We provide 6to4 relay service to everybody in the world.  There is no way
we would offer any tunnel brokering (_using our addresses_) outside our
customers.  6to4 is much more provider-neutral on this aspect, which could 
also be considered as a good thing!

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings