> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On Behalf
> Of Sam Hartman
> Sent: Monday, January 10, 2011 10:28 AM
> To: Klaas Wierenga
> Cc: [email protected]
> Subject: Re: [abfab] draft-lear-abfab-arch-01 posted
> 
> Klaas, thanks so much for your comments.
> I'm responding as an individual not on behalf of the authors as a group.
> 
> I agree that we need to do a better job of discussing federation
selection;
> the current text does not do much about that.  This is definitely a good
> comment and something for us to work on.
> 
> You ask whether we intend to preclude SOAP. I think that's for you to tell
us:
> we've had a WG discussion on this topic and I think we have enough of a
> discussion to approach a consensus. My reading is that we are going to
focus
> on AAA transport but nothing will preclude soap.
> However it would be useful to get a consensus call from the chairs on
this.'

Are we looking at SOAP as a message format or at SOAP as a transport ?  I
don't think that they are necessarily the same thing.


> 
> I agree that the specific protocol flows could be moved later in the
document
> and that would improve things.
> I think it's a little complicated where you want this information: there
is some
> value in having the  diagrams fairly early.

I am in favor of having some diagrams really early in the processes.  I
think this really helps to get understanding setup early.  But I tend to be
a very pictorial person.

> 
> You propose introducing more abstraction: having an abstract discussion of
> the aspects of the system and then discussing specific technology choices.
> Definitely for the application integration, where we've acknowledged in
our
> charter that other options are possible, we should do this. Also, we
should
> separate the abstract discussion of AAA from RADIUS and Diameter.
> However, the use of AAA, EAP, and to some extent SAML is baked in at a
> fairly deep level.
> So, we're not obligated to treat things abstractly.
> We can if it makes the document flow better. However sometimes it is
easier
> to understand one concrete system than a bunch of abstractions.
> So, can we explore how valuable that will actually be?
> 
> 
> 3.3 p14 "GSS-API is specified...": What is the relevance of mentioning the
> sub-operations of the initial context exchange?
> 
> 
> No, not really.
> I thought it might, but it didn't end up being useful.
> 
> 
> 4 p17 It seems that part of this discussion belongs higher up in the
document
> (design goals?) and part is implementation choices
> 
> It's hard to discuss this abstractly. I don't think we have enough
community
> experience for that discussion nor do I think we understand what this
would
> mean abstractly well enough to have this discussion before the discussions
of
> concrete technologies.
> So, I'm sympathetic to the concern, but I'm not sure I know how to do what
> you ask.
> If you want to throw something together or have a chat, we could do that.
> 
> 4.1 p17: explain better what is confusing about the mutual authentication
> terminology
> 
> It's that to some people mutual authentication implies that the client's
> identity is being released.
> In GSS, that's not true.
> 
> 
> 4.2 p18 first paragraph: the sentence starting with "Even when it is
checked"
> is not a well-formed sentence, and I don't understand what it should read.
> 
> Ah, a typeo. s:if the trust:the trust
> 4.2 p19: I don't really understand the whiteboard example, how is the user
> able to validate the content that is being displayed? What if some users
get
> (slightly) different content than others?
> 
> Note here we're talking about a physical electronic LCD whiteboard in a
> room, not some shared conference whiteboard software.Typically a
> presenter can validate the content of their own presentation.  How do we
> deal with attacks that give different users different content?  I claim
that an
> attack that presents different content to people in the same physical room
> looking at the same LCD is outside of the scope of
> ABFAB:-)
> If we make the context clear--that we're talking about a networked LCD--is
> this example more clear?
> 4.2 p19 last paragraph: i think if you explain GSS-API CB you also need to
> explain EAP CB
> 
> I agree that this document needs  to discuss EAP channel binding. We
> propose to do that in the unwritten section 6.1 as part of the deployment
> discussion.
> I guess we could briefly discuss in section 3 as well.

The discussion may also need to include problems with matching names.  If
you have a TLS DNS name and you have some type of Kerbros realm name - they
may not be the same name but still be the same entity.

> 4.3 p19 "Using host based service...": Why does ABFAB need to adopt this
> approach?
> 
> 
> We need to adopt this approach if we want ABFAB to work with IMAP, SMTP,
> LDAP, SSH, NFS or a variety of other protocols that use this naming
approach.
> We could alternatively update all those specifications and update all the
> implementations that use generic frameworks such as SSPI or Cyrus SASL.
> This is a case where following the existing standards makes deploymentand
> implementation easier.
> Can you suggest additional text for the last paragraph of this section to
better
> clarify?
> _______________________________________________
> abfab mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/abfab

_______________________________________________
abfab mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/abfab

Reply via email to