Re: [nznog] [Fwd: [pacnog] IPv4 exhaustion discussionsin Asia Pacific region]
In message <2B310DEB-219F-4F70-9779-F27EDDEE60AA(a)ca.afilias.info>, Joe Abley writes:
On 16-Feb-2007, at 17:00, Nathan Ward wrote:
End-to-end IP connectivity is so passe.
But it's arguably what allowed the Internet to grow and flourish. If the network controls the applications that users can deploy, then [provider catchup or user disappointment or] people will stop dreaming up new applications for users to run, and will instead concentrate on packages they can sell to carriers and ISPs.
Or, perhaps worse, all new applications will have an extra layer to "route around" this problem. The rise of application data over XML over HTTP (over SSL) over TCP over IP hints at what this world might be like. That's not exactly progress. The security folks aren't exactly keen on this either (even without the SSL, it's very hard to "pinhole" inspect what is going on in the midst of large XML payloads, and you don't get much of a service type indicator if it's all XML over HTTP). The "big win" of Skype, and Kazaa before it, is its ability to "route around" the NAT everywhere, firewall everywhere, world and overlay what is basically an end-to-end network again (at least bits of it created on demand). That's not exactly progress either. Some other possibilities have been suggested, such as smarter intermediate boxes (firewalls, NAT), with which one negotiates (as in a protocol to request) a path through to the end box. This is already starting to happen with, eg, STUN, but potentially could be formalised as part of connection setup, and done automatically rather than requiring per-application custom setup. One is left wondering if instead of inventing random ad-hoc "tunnel through intermediate boxes" protocols it may be that things converge on, eg, 6-over-4. At least, eg, application over TCP over IPv6 over IPv4 is still pretty transparent, and doesn't have _that_ much overheard (compared with, eg, XML over HTTP). And there's some eventual migration path which involves reducing layers (eventually when everything is 6-over-4 it makes sense to just get rid of the IPv4 bit). Ewen
participants (1)
-
Ewen McNeill