On 01/08/12 17:40, Andrew Sullivan wrote:
On Wed, Aug 01, 2012 at 04:58:42PM +0200, Wouter Cloetens wrote:
We assume that a DynDNS-style approach simply will never scale for
every IPv6 address in the home, and therefore the home router has to
be authoritative, handling the requests.
If I may ask, why do you make that assumption? I assure you that Dyn
is running some very large zones with high change rates. Now, it
might be that nobody would be willing to pay for such a service at the
rate Dyn would charge to make this profitable, assuming large numbers
of records and a high rate of change. If that's your argument,
however, I'd like to know understand how this is going to be worth
doing for someone else when Dyn already has the sunk investment in
infrastructure.
Internet of Things. Home automation. 5, 10 years from now. 10k nodes
low-power nodes for a site. (Not unrealistic for an office building,
I've been told).
It is sensible to have those nodes visible in the DNS, forward and
reverse, for remote management and other reasons. You basically never,
ever want to type in an IPv6 address.
It is not sensible to have them register themselves to DynDNS, or have
the router register to Dyn for them.
draft-mglt-homenet-naming-delegation and
mglt-homenet-front-end-naming-delegation provide a solution that will
work securely, out of the box, without any end user configuration at
all, and consuming no bandwidth if the home DNS feature is not used. The
end user doesn't even have to register for a DynDNS account.
*That* is what this WG should be about IMO; making things scale to
unimaginable proportions (we are designing for the next 100 years),
simple, without any need for configuration, and without cost.
bfn, Wouter
_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet
