[secdir] Security Review of draft-ietf-behave-nat-behavior-discovery-07

[Dave Cridland <[e-mail]>]

I have re-reviewed this document as part of the security  
directorate's ongoing effort to review all IETF documents being  
processed by the IESG.  These comments were written primarily for the  
benefit of the security area directors.

Original review and response are attached, I omitted to copy the  
secdir mailing list the first time.

For -07, the document addresses the security concerns of the protocol  
extensions involved adequately. In particular:

On Tue May 26 11:28:45 2009, Dave Cridland wrote:
> The Security Considerations section is reasonably complete, as far  
> as  I can tell, however it is not terribly clear that it suggests   
> authentication of the clients (it says "preexisting credentials") -  
> I  think this could be clearer. The description of  
> XOR-RESPONSE-TARGET  also doesn't include this, it's mentioned most  
> clearly in Section 6.1.

This has since been clarified in -07, having been addressed by a  
clear, point-by-point mitigation guide.

Dave.

--- Begin Message ---

I have reviewed this document as part of the security directorate's  
ongoing effort to review all IETF documents being processed by the  
IESG.  These comments were written primarily for the benefit of the  
security area directors.  Document editors and WG chairs should treat  
these comments just like any other last call comments.

(I note there is expected to be a new version coming for this draft).

Security Issues:

The Security Considerations section is reasonably complete, as far as  
I can tell, however it is not terribly clear that it suggests  
authentication of the clients (it says "preexisting credentials") - I  
think this could be clearer. The description of XOR-RESPONSE-TARGET  
also doesn't include this, it's mentioned most clearly in Section 6.1.

General comments:

I have a strong suspicion that this document is Experimental purely  
because it failed to gain sufficient consensus to be Standards-Track.  
It's not clear to me why this is not Informational, or why all the  
extensions described in the document are within the same document.  
I'm dubious that they're all of similar quality.

If there is an experiment here, then it's in the usage of these  
extensions to determine whether, at least in some cases, NAT  
behaviour is sufficiently stable as to be useful, and moreover,  
whether taking advantage of this is practical. The extensions  
themselves clearly seem suitable for discovering whether this is so.

As such, section 2.3 seems somewhat contrived and grasping. This  
isn't to say that the hypothesis being tested is not valid, but the  
experiment, as defined, seems like a matter of form rather than a  
useful test of the hypothesis as outlined.

Editorial Issues:

The use of the term "aprocyphal" is interesting, but conjures up  
connotations that seem to be somewhat self-defeating. Perhaps  
"anecdotal" would be more fitting, or "controversial". (It is this  
evidence, after all, that forms the hypothesis mentioned above, and  
the hypothesis itself is surely not aprocypha).

IANA section requests registration of CHANGE-REQUEST, but this is  
already registered - the registration needs changing, as per section  
6.1, where the situation is detailed more clearly.

Typographical Errors:

Extraneous "}" in section 9.4.

--- End Message ---

--- Begin Message ---

Dave,

Thanks for the feedback.   Responses inline.


On Tue, May 26, 2009 at 6:28 AM, Dave Cridland<[e-mail]> wrote:
> I have reviewed this document as part of the security directorate's ongoing
> effort to review all IETF documents being processed by the IESG.  These
> comments were written primarily for the benefit of the security area
> directors.  Document editors and WG chairs should treat these comments just
> like any other last call comments.
>
> (I note there is expected to be a new version coming for this draft).
>
> Security Issues:
>
> The Security Considerations section is reasonably complete, as far as I can
> tell, however it is not terribly clear that it suggests authentication of
> the clients (it says "preexisting credentials") - I think this could be
> clearer. The description of XOR-RESPONSE-TARGET also doesn't include this,
> it's mentioned most clearly in Section 6.1.

I made a lot of edits around XOR-RESPONSE-TARGET, in the process I
actually removed the phrase "pre-existing credentials", just leaving
it as authenticated.  However, the term comes from 5389, which only
defines mechanisms for using "pre-existing credentials" for
authentication, meaning credentials that are obtained through a
mechanism outside STUN itself.

>
> General comments:
>
> I have a strong suspicion that this document is Experimental purely because
> it failed to gain sufficient consensus to be Standards-Track. It's not clear
> to me why this is not Informational, or why all the extensions described in
> the document are within the same document. I'm dubious that they're all of
> similar quality.
>
> If there is an experiment here, then it's in the usage of these extensions
> to determine whether, at least in some cases, NAT behaviour is sufficiently
> stable as to be useful, and moreover, whether taking advantage of this is
> practical. The extensions themselves clearly seem suitable for discovering
> whether this is so.
>
> As such, section 2.3 seems somewhat contrived and grasping. This isn't to
> say that the hypothesis being tested is not valid, but the experiment, as
> defined, seems like a matter of form rather than a useful test of the
> hypothesis as outlined.

Section 2.3 doesn't describe an experiment, it describes conditions
for experimental success.  Section 2.2 has been greatly expanded and
now describes in much more detail how such an application might work.
However, it's not the only application that could satisfy the
conditions.

>
> Editorial Issues:
>
> The use of the term "aprocyphal" is interesting, but conjures up
> connotations that seem to be somewhat self-defeating. Perhaps "anecdotal"
> would be more fitting, or "controversial". (It is this evidence, after all,
> that forms the hypothesis mentioned above, and the hypothesis itself is
> surely not aprocypha).

Another reviewer also brought this up.  Technically, "aprocryphal"
could be interpreted as being contrary to IETF dogma, but I think
you're right that anecdotal is clearer here.

>
> IANA section requests registration of CHANGE-REQUEST, but this is already
> registered - the registration needs changing, as per section 6.1, where the
> situation is detailed more clearly.
>

Numerous attempts by myself and others to determine the right way to
handle this have indicated that this is the appropriate way to handle
this, as the original registration has been obsoleted.  But I'm always
open for advice as to the best way to handle it.

Bruce

--- End Message ---