[e2e] The Internet is too direct (was: [e2e] using p2p overlays to overcome recursive NATs/realms)

[Michael Welzl]

Christian,

I agree 100%. That's what it all seems to be about these days. And I
don't think it's wrong ... it seems to be kind of natural, and it may
even be healthy.



On Tue, 2002-02-12 at 12:25, Christian Tschudin wrote:
> Jon's initial posting was on NATs and overlays, which I think is an
> instance of a more general theme: "indirection".
> 
> The Internet is too direct and IP takes too many shortcuts. In fact,
> several "enhancements" are reactions to this too direct a style:
> 
> - p2p overlays are a reaction to the lack of methods to influence
>   routing, run your own address space
> 
> - you ask for a web page and get the content from another place

in this contet, these two "enhancements" remind me of Van Jacobson's
keynote speech at last year's sigcomm... it's about time we got rid of
the circuits.


> 
> - ping is too direct, smurf likes this, firewalls fight against it
> 
> - the "end address" is too direct, mobile IP had to invent its own
>   redirection
> 
> - peeking at an IP packet's src field to learn about the sender
>   is a problematic shortcut that skips a resolution step
> 
> - directness also has become an evolution problem: turning the
>   standardization knob here alters things allover there - too direct
>   coupling
> 
> The trend I see is towards requesting and adding more indirection,
> for routing, security, economic, transition whatever reason. Not
> all of these attempts will be as transparent as the AS thing (another
> example of added indirection).
> 
> And the architectural consequences? In the long run
> 
> - IP becomes an access protocol and emulation target, overlays
>   creating partial and transient clouds of emulated directness
> 
> - IP "end addresses" become "start addresses" which the network
>   has to resolve for you. NAT boxes are just a half step in this
>   direction.

I consider Multihoming in SCTP another, maybe smaller but nicer step in
this direction. It's an example of how things that are developed as
overlays descend in the protocol stack later on.


> 
> christian.
> 
> 
> On Fri, 8 Feb 2002, Jon Crowcroft wrote:
> 
> > 
> > so the problem with most the proposed solutions to workign around nats
> > is that they really assume there are only 2 realms - 
> > the great unwashed internet, and the poor deprived natted user. 
> > 
> > the real situation is that packets might traverse multiple natted realms (c.f. realm
> > specific ip) - in this scenario, discovering the mapping involves discovering a path of
> > several mappings-
> > 
> > soluton might be to start a p2p service, which propgates mappings - take the ideas from
> > stun, turn, rsip etc, and use them repeatedly...where multicast is available use it
> > 
> > where one can infer the infernal internal algorithm used by a nat, use it. 
> > 
> > if the p2p service thus built (we might call it an InterNAT) has either dynamic DNS update, or
> > uses ipv6 itself, then to provide global reachability is quite simple...
> > 
> >  cheers
> > 
> >    jon
> 
> ---
> Christian Tschudin, Uppsala University, IT Dept., Box 325
> S-75105 Uppsala, Sweden. http://www.docs.uu.se/~tschudin/
> 


Cheers,
Michael