On Tue, Oct 5, 2010 at 8:16 PM, Noel Chiappa
<jnc(_at_)mercury(_dot_)lcs(_dot_)mit(_dot_)edu>wrote:
> From: RJ Atkinson <rja(_dot_)lists(_at_)gmail(_dot_)com>
> It seems so incredibly unlikely that end-to-end connectivity (i.e.
> without NAT, NAPT, or other middleboxes) is going to increase in
future.
Indeed. It seems that the likelihood of IPv6 being used ubiquitously to
provide end-end IPv6-IPv6 connectivity, as originally envisioned, is fairly
small; instead, it seems we are headed for a future of various kinds of
lash-ups (e.g. the scenario you posited with content providers, or with
IPv6
being used between the cable modem and some sort of CGN IPv6/IPv4 NAT, with
IPv4 on the other pieces of the path, as one large ISP has proposed).
The interesting question, of course, is whether (and if so, when) the IETF
will deign to notice this reality - or will it continue to prefer to stick
its collective fingers in its ears and keep going 'neener-neener-neener'.
Application designers who produce designs that rely on IP addresses being
end-to-end are going to find their work fails.
Since the one legacy protocol that has a dependency on IP address constancy
is FTP, it would seem to me to be much easier to upgrade FTP to remove the
dependency than to try to control the network.
Since FTP is layered on Telnet (no really, it is) it would seem that the
more sensible approach would be to standardize the file handling extensions
to SSH and move FTP to historic status.
At the moment we have a situation where everything layers over HTTP or HTTPS
because they provide NAT passthrough.
--
Website: http://hallambaker.com/
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf