Re: NAT traversal and its relation to IPv6 [RE: Comments on draft-tsirtsis-dsmip-problem-01.txt]

[Pekka Savola <pekkas@netcore.fi>]

On Thu, 4 Sep 2003, Alexandru Petrescu wrote:
> Pekka Savola wrote:
> >> With respect to dual-stacks, the thing I'd like to have is the
> >> Mobile IPv6 MN-HA tunnel to be a v6-in-v4 tunnel instead of
> >> v6-in-v6
> > 
> > My concern is this: we already have about half a dozen IPv4 NAT
> > traversal techniques (IPsec, MIPv4, STUN, TURN, a number of others)
> > -- and some have been proposed to allow IPv6-in-IPv4 tunneling to
> > also traverse NAT (none taken up by this WG, at the moment, though.).
> 
> Good to have a rich set of proposals, selecting is easier than building.

That's one perspective, with a set of tradeoffs.  Some (many?) people also 
argue otherwise: with too much choice, the user if faced with a dilemma on 
what to choose, and the vendor is faced with the dilemma on what to 
implement (so it has to implement all), etc.
 
> > So you want to implement and specify one for dual-stack Mobile IPv6 
> > *too*?
> 
> No, I want Mobile IPv6 to not require its MN-HA tunnel to be v6-in-v6
> but to require it to be either v6-in-v6 or v6-in-v4.

But because you'd like that to work also where v4 NAT is done, a simple 
v6-in-v4 tunnel would not be enough -- it would have to be able to handle 
NAT traversal -- right?
 
> > My _personal_ two top priorities in this context:
> > 
> > 1) make sure NAT traversal stays an _IPv4_ problem.
> 
> I've heard this concern before, so I guess it is right.  Also, there 
> were concerns expressed that NAT's are unqualifiable and unclassifiable 
> beasts, so one can not really find a common means to drill through all 
> of them.  Which is perfectly right to a certain extent.

Well, I think this depends *very* much on how big holes you want to drill.  
Bidirectional tunneling?  No problem.  Drilling through from pretty much 
everywhere?  Certainly there are issues.

[...]
> Because I doubt that, I'm bringing back the mobility aspect: mobility
> systems are deployed only in a limited set of kinds of systems (say GPRS
> and hotspots, but there are others), cover only a limited set of types 
> of NAT; for those 3-5 types of NAT's maybe a common NAT traversal 
> mechanism can be found.

I'm not sure if we can make that kind of assumptions about Mobile IP 
deployment, but if we can, fine -- it's still a lot.

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