Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Re: [dnsext] a note on the discussion about DNSCurve and DNSSEC

0 views
Skip to first unread message

Ólafur Guðmundsson

unread,
Sep 5, 2008, 9:25:00 AM9/5/08
to
Thank you Roy, for your excellent message.

WG: the chairs think this is an excellent path to follow.

The chairs encourage the creators of DNScurve to bring any documentation or
requirements to the working group for either formal or informal discussion.

Olafur & Andrew


At 06:04 05/09/2008, Roy Arends wrote:
>I think it is safe to assume that though DNSSEC and DNSCurve have some
>overlap in functionality, both try to solve different problems. They are
>not mutually exclusive, there does not need to be a choice. There is no
>need, IMHO, to go into polarizing debates about functionality, or lack of
>functionality. May I suggest that we wait for an internet-draft, some
>running code for DNSCurve and some rough consensus? Personally, I would
>appreciate it if DNSCurve could stand on its own, without clinging to
>perceived deficiencies of DNSSEC.
>
>After all, this mailing list is an IETF based list, open to all who are
>willing to contribute (positively). It would be so much more productive if
>we can test interoperability between various DNSCurve implementations;
>Build those implementations on a sound specification, and that the
>specification comes in a form we are comfortable with: an internet-draft.
>
>Such an internet-draft (or set of internet drafts) would most likely need
>to include a problem statement, a set of requirements, an introduction to
>the proposed solution, etc.
>
>Without such, we're wearily debating in circles, stalling process on
>standardization of these solutions. Lets not continue this modus operandi.
>
>Regards,
>
>Roy Arends
>Nominet


--
to unsubscribe send a message to namedroppe...@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/namedroppers/>

bman...@vacation.karoshi.com

unread,
Sep 5, 2008, 11:10:28 AM9/5/08
to
On Fri, Sep 05, 2008 at 07:47:15AM -0700, Eric Rescorla wrote:
> At Fri, 05 Sep 2008 09:25:00 -0400,

> Slafur Gupmundsson /DNSEXT chair wrote:
> >
> > The chairs encourage the creators of DNScurve to bring any documentation or
> > requirements to the working group for either formal or informal discussion.
>
> I think requirements are probably what's most appropriate here.
>
> At a high level, there are two general architectural approaches that
> can be taken to this problem: object security and channel security,
> with DNSSEC being an example of the former and DNSCurve being an
> example of the latter.
>
> If there's general agreement that the IETF should pursue a channel
> security approach as well, then it's reasonable to discuss what
> the properties of that channel security protocol ought to be and
> how best to fill that gap. We already have a lot of experience
> with designing such protocols and it's not at this point clear
> to me that we need an entirely new one.
>
> -Ekr

given that the IETF did TSGI and SIG(0) and TSIG and SIG(0) are
both methods for channel security(#), it seems reasonable to
consider other methods for channel security within the IETF.

# can we use the more accurate wording here? what we are really after
is integrity of the channel, not any of the other attributes of the "security"
umbrella.


--bill

Francis Dupont

unread,
Sep 5, 2008, 11:49:35 AM9/5/08
to
In your previous mail you wrote:

If there's general agreement that the IETF should pursue a channel
security approach as well, then it's reasonable to discuss what
the properties of that channel security protocol ought to be and
how best to fill that gap. We already have a lot of experience
with designing such protocols and it's not at this point clear
to me that we need an entirely new one.

=> and please note we already have TSIG and its associated protocols
(TKEY, RRSIG(0), ...).

Regards

Francis...@fdupont.fr

PS: as I wrote because of the caching & no-trust infrastructure of the DNS
what we need first is what you call the "object security".

Paul Hoffman

unread,
Sep 5, 2008, 12:05:48 PM9/5/08
to
At 9:02 AM -0500 9/5/08, Jon A. Solworth wrote:
>> There is no need, IMHO, to go into polarizing debates about
>>functionality, or lack of functionality.
> Why should a technical discussion be polarizing?

I take it you are new to the IETF, then. :-) I'm sure Dan can point
you to some archives of technical discussions with which he is
familiar that could be described as "polarized".

>>May I suggest that we wait for an internet-draft, some running code
>>for DNSCurve and some rough consensus? Personally, I would
>>appreciate it if DNSCurve could stand on its own, without clinging
>>to perceived deficiencies of DNSSEC.
>>

> Perhaps I'm misunderstanding what you intended, but it sounds
>like your suggestion is that wait to discuss that until we have consensus
>(and running code).

It doesn't sound like that at all. The request for an Internet Draft
is so that the full proposal for the protocol can be considered. At
this point, we don't even have a protocol diagram, much less a
description of what information is carried on the wire.

> How are we going to get consensus without discussion?

A more apropos question is "how are we going to have a useful
discussion without a clear and complete first draft of the proposal"?

> Would it not be good to understand a basis for evaluation before the
> code exists?

Yes.

--Paul Hoffman, Director
--VPN Consortium

Paul Hoffman

unread,
Sep 5, 2008, 12:54:56 PM9/5/08
to
At 11:46 AM -0500 9/5/08, Jon A. Solworth wrote:
> I think that DJB wants to follow a model of implementation
>precedes standardization;
>given this model a standards document would be premature. But the
>suggestion of
>an informal presentation to IETF is interesting.

Just to be clear: I didn't suggest that. I agreed with the others who
have said than an Internet Draft describing the protocol is the
appropriate way forwards if you want DNSCurve to be discussed in the
IETF.

Paul Hoffman

unread,
Sep 5, 2008, 1:05:15 PM9/5/08
to
At 5:49 PM +0200 9/5/08, Francis Dupont wrote:
>=> and please note we already have TSIG and its associated protocols
>(TKEY, RRSIG(0), ...).

To be fair, it is clear from the discussion that DNSCurve describes a
way to scalably produce and distribute the keying material needed for
something like TSIG. TSIG without something like DNSCurve is like
IPsec without IKE

0 new messages