Cable transition [RE: Opportunistic Tunneling]

[Pekka Savola <pekkas@netcore.fi>]

I think the most interesting discussion is what you say about cable 
transition.. but I'm responding to a few other issues for 
clarification as well.

On Mon, 23 Feb 2004, Tony Hain wrote:
> Pekka Savola wrote:
> > No, ISATAP is not really applicable AFAICS.  Remember that the topic
> > here is "opportunistic tunneling" -- with a requirement that the
> > tunneling works independent of ISPs, automatically.  ISATAP requires
> > ISATAP router and set-up.
> 
> Only to get out of the routing domain. ISATAP end systems will use
> opportunistic tunneling directly because ND is not required to acquire the
> IPv4 (link layer) address. 

Yep - ISATAP includes the concept of intra-site opportunism.  But as 
it is not opportunistic in inter-domain (which was the point here, the 
ISPs not offering service), it does not seem too interesting.

> As Bob & Christian have said, the IETF is not in a position to recommend
> economic models. 

I have to disagree with the way people have framed the IETF position, 
while I agree with your statement above.  But I don't think this is 
something to continue on this thread.

> Entire industries like cable ISPs will have economic
> drivers (like the cost/operational complexity of whole-sale replacement of a
> cable modem plant) that will make technologies like 6to4 & Teredo much more
> attractive than they would be to a campus or transit ISP. Operating 6to4 or
> Teredo relays as an integral part of a cable infrastructure provides a very
> simple deployment path. At most the IETF might document that the IPv4 side
> of such relays should be restricted to local customers, but even that is
> really none of the IETF's business.

Is there anything fundamentally unworkable in the model where the 
cable modem users would use a tool such as ISATAP, STEP or TSP to 
connect to the tunnel server of the cable modem ISP for their 
connectivity?

This seems to fail in three cases only:
 1) if the cable ISP doesn't want to provide IPv6 in the first place 
(not even providing 6to4 relay or whatever).  Based on your comment, 
this doesn't seem a bad concern.

 2) if the traffic between the users at the same ISP grows so high 
that a tunnel server (or multiple servers) cannot carry all the 
traffic, i.e., path optimization would be required.  Personally, I see 
this as an indication that IPv6 would be in serious use and the cable 
ISP would want to start offering native IPv6.  So, this would only 
seem to be the case, if the cable ISP wanted to avoid transition to 
native IPv6 altogether, sticking to 6to4/Teredo for forever .. which 
is not what we want.

 3) if the mechanism (ISATAP, STEP, TSP, whatever) is not implemented 
widely enough in the host software so that it's unavailable to the 
users.

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