It is technically possible for the United States and Europe to switch from
driving on the wrong side of the road to the left. That does not mean that
proposing to do so is anything other than idiotic.
What I said was that these proposals amount to distinction without a
difference. The switching costs are huge, the end result indistinguishable
except to the extent that you have multiplied the number of failure modes
drastically.
-----Original Message-----
From: Emin Gun Sirer
[mailto:egs(_at_)systems(_dot_)cs(_dot_)cornell(_dot_)edu]
Sent: Tuesday, November 28, 2006 7:32 PM
To: Stephane Bortzmeyer
Cc: Patrick Vande Walle; ietf(_at_)ietf(_dot_)org
Subject: Re: Something better than DNS?
Stephane & Phillip,
I'm thinking of writing a short report that summarizes the
invaluable discussion here and beefing up the system sketch.
I think we now agree that it is possible to have multiple
operators manage names in a single, shared namespace without
recourse to a centralized super-registry.
Do you think it fits well in Hallam-Baker, Phillip's
"logical registry"
model?
Yes, the registrars together implement a logical registry
without any centralized component.
* how does it scale?
The protocol I sketched was for just two parties; it can be
generalized, but let's take that discussion offline. It's
certainly possible to do it, but something that'll work fast
and with participants that will trigger timeouts might
require quorums, a slightly different approach.
* DNS registration is not binding a name to an IP! (See
Edward Lewis'
good description). It is binding a name to an entity which is often
fuzzily defined (see whois' output). How do you address this?
One can use a protocol like that to just serialize the events, i.e.
establish order, regardless of what the events that comprise
a registration are.
What you absolutely cannot get away from is a mechanism to
ensure joint
action.
True. The question was whether the mechanism necessarily had
a centralized component.
the point of the exercies here is to meet the needs of the Internet
users, not solve cute academic puzzles for the sake of arbitrary
ideological goals.
For one, I am totally agnostic on this issue. It's possible
and maybe even practical to delegate a space to multiple
operators without a super-registry, and neither I nor the
workings of CoDoNS care whether we do that or not.
As an Internet user, I wonder about two things in the long term:
- why is it so expensive to register a name?
- what can we do to keep SiteFinderJr from happening?
In the short term, I think we have bigger issues to worry about (e.g.
current DNS's vulnerability to denial-of-service attacks).
Gun.
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf