Re: [secdir] [Netconf] review of draft-ietf-netconf-call-home-11

[Kent Watsen <[e-mail]>]

Hi Simon,


Regarding your two comments:

1. Are non-certificate-based TLS out of scope for NETCONF/RESTCONF?

Yes or, more specifically, both NETCONF over TLS and and RESTCONF restrict TLS to certificates.  The relevant text is in RFC 7589, Section 1 and draft-ietf-netconf-restconf-07, Section 2.2.   Do you think there should be a note in this draft stating this as well?

2. Section 2 says 'The term "NETCONF/RESTCONF client" can refer to the [RFC6241], Section 1.1 "client".'

Does Benoit's suggestion to replace "can refer" with "refers" satisfy the language vagueness issue?   - And also, are you satisfied with having the single reference to RFC6241, since RESTCONF pulls the term from RFC6241 as well?


Thanks
Kent


From: Benoit Claise <[e-mail]<mailto:[e-mail]>>
Date: Monday, October 12, 2015 at 3:50 AM
To: Simon Josefsson <[e-mail]<mailto:[e-mail]>>, "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>, "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>, "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>, "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: Re: [Netconf] review of draft-ietf-netconf-call-home-11

Hi Simon,

Thanks for your review.
One comment below.

Hi.

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 believe the document is ready.

One main security concern is the reversal of roles that this document
introduce, but letting TCP clients act as TLS/SSH servers, and vice
versa, is not unheard of.  As long as proper peer authentication is
performed, and other parts of the security protocols are properly
performed, I see no fundamental problem with this.  I'm sure some
implementations will need to be tweaked to deal with this, and
terminology might confusing at times.  The 'Security Considerations'
section does a good job discussing this, and some other issues too.

Two minor questions:

1) Are non-certificate-based TLS out of scope for NETCONF/RESTCONF?  I
see no discussion of it in this draft, and text in the document
implicitly assumes host keys (SSH) or certificates (TLS) are used.
Think about TLS-PSK for example, which seems like a relevant idea for
embedded devices.  This may not be the document to adress this, but if
there is work towards that goal already, it might be useful to align
this document with that.

2) Section 2 says 'The term "NETCONF/RESTCONF client" can refer to the
[RFC6241], Section 1.1 "client".'.  Shouldn't this say the term may
refer to a RESTCONF client and reference draft-ietf-netconf-restconf?
Or is that not intentional?  The use of the word 'can' make this text
vague to me.  The previous section (1.5) says that 'NETCONF/RESTCONF' is
an abbrevation for 'the NETCONF or the RESTCONF'.  The same comment
applies to section 3.  Maybe this is a misunderstanding on my side, but
the text confused me so it may be useful to resolve.

I mentioned this point in my AD review (http://www.ietf.org/mail-archive/web/netconf/current/msg10535.html).
I now realize that I made a typo in my email: NETCONF/RESTCONF client" can refers to the RFC 6241 section 1.1 "client"
This should read:  "NETCONF/RESTCONF client" refers to the RFC 6241 section 1.1 "client"
Ditto for NETCONF/RESTCONF server.
This takes care of the confusing "can" word. Thanks for spotting that.

Regarding the draft-ietf-netconf-restconf reference now.
https://tools.ietf.org/html/draft-ietf-netconf-restconf-07#section-1.4.1 refers to RFC 6241 for the client.
So I believe that we're good with a single reference to RFC 6241 in this draft.

Regards, Benoit

Thanks,
/Simon