Re: [Gen-art] [Perc] Genart last call review of draft-ietf-perc-srtp-ekt-diet-09

[Richard Barnes <[e-mail]>]

Hi Christer,

Sorry, just now seeing this.  Responses inline.  PR here:

https://github.com/ietf/perc-wg/pull/165

--RLB

On Sat, Oct 20, 2018 at 6:06 PM Christer Holmberg <
[e-mail]> wrote:

> Reviewer: Christer Holmberg
> Review result: Ready with Issues
>
> I am the assigned Gen-ART reviewer for this draft. The General Area
> Review Team (Gen-ART) reviews all IETF documents being processed
> by the IESG for the IETF Chair.  Please treat these comments just
> like any other last call comments.
>
> For more information, please see the FAQ at
>
> <https://trac.ietf.org/trac/gen/wiki/GenArtfaq>.
>
> Document: draft-ietf-perc-srtp-ekt-diet-09
> Reviewer: Christer Holmberg
> Review Date: 2018-10-20
> IETF LC End Date: 2018-11-01
> IESG Telechat date: Not scheduled for a telechat
>
> Summary: The document is well written, and easy to read. But, I think some
> things still need to be clarified. I also have some editorial modification
> suggestions.
>
> Major issues:
> --------------
>
> Q1_1:
>
> The text in section 1 says that EKT removes the need for co-ordinating
> SSRCs,
> and that an endpoint can choose whatever value it wants.
>
> However, I can't find any explanation on how EKT removes that need.
>

The answer is toward the end of the introduction: "EKT does not control the
manner in which the SSRC is generated..." -- it just makes the distribution
of security information easier.  I extended that paragraph to clarify.



> Q1_3:
>
> The text in section 1 says that EKT is not intended to replace key
> establishment mechanisms, but to "be used in conjunction".
>
> However, there is no description on how that works in conjunction with
> existing
> mechanisms (e.g., together with SDP Offer/Answer), and whether existing
> mechanisms need to be modified in order to work in conjunction with EKT.
>
> Section 5.3 does say that the SDP O/A exchange is not affected. If that is
> true, then you need to describe how SSRC values etc signalled in SDP
> relate to
> values signalled using EKT, what happens if there are mismatches etc.
>

This document only defines one such augmentation -- the integration with
DTLS-SRTP.  I extended the relevant part of the introduction to say this.



> Minor issues:
> --------------
>
> Q1_2:
>
> The text in section 1 says:
>
>    "However, there are several cases in which conventional signaling
> systems
>    cannot easily provide all of the coordination required."
>
> I think some examples would be useful.
>

Disagree.


> Q1_4:
>
> The text in section 1 says that providing SSRCs etc using signaling systems
> cause layer violations.
>
> Is this layer violation described somewhere? If so, I think a reference
> would
> be useful.
>

I just deleted this sentence.  It wasn't adding much.



> Q4-2_1:
>
> The text in section 4.2 says:
>
>    "All of the received EKT parameter sets SHOULD be stored by all of the
>    participants in an SRTP session, for use in processing inbound SRTP
>    and SRTCP traffic."
>
> What is the reason for SHOULD, instead of MUST? What happens if an endpoint
> does NOT store the EKT parameter sets?
>

I think the SHOULD here is to allow, e.g., cache management.  Clarified the
consequences.



> Q4-2-1_2:
>
> The text in section 4.2.1 says:
>
>    "Outbound packets SHOULD continue to use the old SRTP Master Key for
>    250 ms after sending any new key."
>
> What is the reason for SHOULD, instead of MUST?
>

I don't think this should be a MUST.  For example, how accurately are you
going to measure compliance?  If the change-over is 10ms short 10% of the
time, is the endpoint out of compliance. Plus, there's minimal interop
impact; this just helps things go smoothly.



> Q4-3_2:
>
> The text in section 4.3 says:
>
>    "Section 4.2.1 recommends that SRTP senders continue using an old key
>    for some time after sending a new key in an EKT tag."
>
> The text in section 4.2.1 contains a SHOULD, so it is more than a
> recommendation.
>

According to RFC 2119, a SHOULD is precisely a recommendation:

"""
3. SHOULD   This word, or the adjective "RECOMMENDED", mean that there
   may exist valid reasons in particular circumstances to ignore a
   particular item, but the full implications must be understood and
   carefully weighed before choosing a different course.
"""

https://tools.ietf.org/html/rfc2119#section-3



> Nits/editorial comments:
> --------------------------
>
> Q1_5:
>
> I think Section 1 should also indicate that EKT is an DTLS extension,
> similar
> to the Abstract. Otherwise, when you say that EKT is a way to distribute
> information, it is unclear why EKT doesn't violate layers.
>

Done.



> I think that Section 2 (Overview) could be combined with Section 1
> (Introduction). Introduction sections in RFCs typically provide both
> background, problem statement and an overview of the mechanism - and
> sometimes
> also the document structure.
>

Disagree.


> Q4_1:
>
> The text in section 4.1 says:
>
>    "EKT MUST NOT be used in conjunction with SRTP's MKI (Master Key
>    Identifier) or with SRTP's <From, To> [RFC3711], as those SRTP
>    features duplicate some of the functions of EKT.  Senders MUST NOT
>    include MKI when using EKT.  Receivers SHOULD simply ignore any MKI
>    field received if EKT is in use."
>
> I suggest to put this text in a dedicated Applicability section.
>

Disagree.


> Q4-1_1:
>
> The text in section 4.1 says:
>
>    "Together, these data elements are called an EKT parameter set.  Each
>    distinct EKT parameter set that is used MUST be associated with a
>    distinct SPI value to avoid ambiguity."
>
> I suggest to defined "EKT parameter set" in the same way as the other
> terminology. I.e.,
>
>   "EKT parameter set: The parameters indicated by the SPI"
>
> ....or something like that.
>

There's not a terminology section, so I didn't do exactly what you say.
But I pulled this out into a separate section so that it doesn't interrupt
the flow, and clarified that the SPI-params association is set up by the
DTLS-SRTP extension.


Q4-2-1_1:
>
> For the section names of sections 4.2.1 and 4.2.2, I suggest to talk about
> sending/transmitting and receiving instead of inbound and outbound.
>

Disagree


> Q4-3_1:
>
> In section 4.3, I  think the name of the section ("Implementation Notes")
> is a
> little confusing. Also, is there a reason why the text is not in section
> 4.2.1
> and/or 4.2.2?
>

Just folded this into section 4.2.2.



> Q4-4_1:
>
> Sections 4.4 and 4.4.1 have the same section names. Please change one of
> them
> (e.g., change 4.4.1 to "Default Cipher").
>

Changed to "AES Key Wrap"



> Q4-6_1:
>
> I think the text in section 4.6 should be placed in the Applicability
> section I
> suggested earlier (Q4_1).
>

Merged it into section 4.



> Q5_1:
>
> In section 5, I  suggest to change the section name to "DTLS-SRTP
> Considerations".
>

Disagree.


>
>
> _______________________________________________
> Perc mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/perc
>