> > http://www.cavebear.com/cavebear/growl/
>
> Well and succinctly put.
Thanks.
> Cooperation among the various root operators (or lack thereof) would likely
> determine the best working model.
When you say "various root operators" do you mean the server operators or
the operators of the each of the groups of servers, each of which
constitutes what I've been trying to call a "root system"?
There certainly needs to be some firm cooperation between the operators of
root servers who belong to a given root system.
But as far as inter-root system cooperation goes -- I don't see the need
for there to be any beyond adhering to a common protocol standard. I see
enlightened self interest as a force that will cause there to be no
net-instability causing practices.
By-the-way, I forget to mention the "value added service" that I didn't
put into the URL mentioned above -- that is that when there are multiple
root systems, one can subscribe to one that will vector you to TLD servers
which are in parts of the net that are topologically close to where you
are.
This can have significant benefits. DNS queries won't end up travelling
so far across the net (this saves potentially mongo on long-haul
bandwidth), response times ought to be shorter (although long haul delays
on today's net aren't all that bad), and the traffic would transit fewer
exchange/peering points, one of the the prime places where internet packet
loss occurs.
--karl--
