Re: My two cents on TLS mandatory ciphers

[Jeff Williams <jwkckid1@xxxxxxxxxxxxx>]

Ned and all,

Ned Smith wrote:
> 
> At 03:03 PM 7/14/97 +0100, Jeff Williams wrote:
> >somebody wrote:
> >>
> >> On Mon, 14 Jul 1997, David P. Kemp wrote:
> >> > I have seen only two rationales for using MUST instead of SHOULD:
> >> >
> >> > 1) mandatory is necessary to ensure interoperability.
> >> > 2) the IESG remanded the document for lack of a MUST.
> >> >
> >> > Neither of those reasons is persuasive*.  If you want to convince the
> >> > working group, you're going to have to come up with a better set of
> >> > reasons than those.
> >>
> >> Perhaps not to persuasive to you, but both are substantive points.  I have
> >> countered every objection to a mandatory cipher suite with a specific
> >> counter-proposal.
> >>
> >> Nobody has countered point (1), except by reference to an unexplained
> >> proprietary mechanism which is not required by TLS and thus not relevant.
> >> Nobody has countered point (2), which must be addressed if we want a TLS
> >> standard.
> >
> >  Well I cannot say much in response to "2".  But I have countered
> 
> Clearly the IETF cannot create a standard that dictates the security policy
> of the users of that protocol! It behooves us to get clarification from the
> IESG regarding response #2. I find it hard to believe the IESG would
> knowingly dictate/mandate site security policy. We must stick to good
> design priciples of separating policy from mechanism. Tim's proposal
> identifying SHOULDs instead of MUSTs is sensitive to this priciple while
> maximizing the potential for interorability. A protocol that forces me to
> use an unsafe protocol is itself an attack. We cannot mandate a perfect
> policy since threat follows value.

  I agree with this analisys completly.  FOrcing any cipher suite as
a manditory default in in and of itself an attack or at least one in the
making.
This is not good design as I have said myself on several occasions, and
indeed not necessary if the design is done properly.
> 
> At the last IETF meeting it was decided that there would be a companion
> working group to define ciphersuites and assign name space. This is still
> the best way to achieve interoperability between TLS endpoints. IMO this
> thread of discussion rightly belongs in that new working group. Historical
> precedent in the IETF might require protocols to interoperatate, but
> security in the IETF is largely unprecedented.

  Good point.
> 
> Regards,
> Ned Smith
> Ned Smith                 Intel Architecture Labs
> JF2-74  2111 N.E. 25th Ave.  Hillsboro, OR. 97124
> Ph: 503.264.2692 Fax: x1805  Mailto:nsmith@xxxxxxxxxxxxxxx
> Http://www.intel.com/ial/security/index.htm
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Regards,
-- 
Jeffrey A. Williams
DIR. Internet Network Eng/SR. Java Development Eng.
Information Eng. Group. IEG. INC. 
Phone :913-294-2375 (v-office)
E-Mail jwkckid1@xxxxxxxxxxxxx