Last Call Review of draft-ietf-trill-ia-appsubtlv-08
review-ietf-trill-ia-appsubtlv-08-genart-lc-kyzivat-2016-06-27-00

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
<http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.

Document: draft-ietf-trill-ia-appsubtlv
Reviewer: Paul Kyzivat
Review Date: 2016-06-27
IETF LC End Date: 2016-06-28
IESG Telechat date: 2016-07-07

Summary:

This draft is on the right track but has open issues, described in the
review.

This is a well written document. I was generally able to follow it even
though I know nothing about the subject.

Issues:

Major: 0
Minor: 7
Nits:  2

(1) MINOR: (Section 2)

"Addr Sets End" is described as follows:

    o  Addr Sets End: The unsigned integer offset of the byte, within the
       IA APPsub-TLV value part, of the last byte of the last Address
       Set. This will be the byte just before the first sub-sub-TLV if
       any sub-sub-TLVs are present ...

But the remaining text of this section, and the examples, imply that
this is really the length of the leading portion of this TLV ending with
the last Address Set. The programmer in me says these differ by one, and
that the implied definition is the reasonable one, while the action
definition, and the name used to identify it, are wrong.

I expect it would be difficult at this point to rename this field, but
at least the definition can be rewritten to be consistent with the
intended usage.

(2) MINOR: (Section 5.1)

Normally I would expect this section to request IANA to assign new
values from the AFN table for OUI...RBridge Port ID. However it is
worded as "IANA has allocated". Perhaps this is because they have
already been (pre)allocated. I have no problem with that if IANA is OK
with it.

But IMO the references to IPv4...64-bit MAC are gratuitous and
inappropriate in an IANA Considerations section. If it is desired to
include a list of "useful" AFN values then that belongs in some other
portion of the document.

(3) MINOR: (Section 5.1)

The "new" values here (OUI, MAC/24, MAC/40, IPv6/64) give "This
document" as their reference. But anyone consulting the IANA registry
and following it to this document would have difficulty finding any
*definition* of these things.

Section 6 discusses some operational issues with them, but at best
implies a definition. (RFC7042 might be considered a definition of OUI,
though it doesn't seem to say how big it would be.)

I think what is needed are explicit definitions of all of these,
including their widths. (In order to provide enough bits to complete a
MAC/24 it must be at least 24 bits wide, but that would be bigger than
needed for a MAC/40. So I guess it must be at least 24 bits, and when
used to expand a MAC/24 or MAC/40 an appropriate number of its high
order bits are used.)

It would be good for there to be a section, appearing in the TOC, for
each of these so that someone coming here from the IANA registry will
easily be able to find the definition.

(4) MINOR: (Section 5.2)

This section defines a new registry with Expert Review as the procedure
for approving new entries. What I don't see is any guidance to the
expert on appropriate criteria to use to judge suitability of new
entries. Without any guidance, relying on the whim of the expert can
lead to variable, and perhaps biased, results.

It would be good to give guidance on: what sorts of document reference
are acceptable, what information needs to be included in the reference
document, whether "special" values may be requested (versus just
assignment in order requests are received), and the sorts of properties
that are appropriate.

(5) MINOR: (Section 6)

This section talks about the handling of OUI and IPv6/64 when they
appear in a Fixed Address sub-sub-TLV. It says nothing about their
meaning if these appear elsewhere, such as in a Template. I presume this
kind of usage is nonsense, but it would be better to explicitly state it.

(6) MINOR: (Section 6)

The description of IPv6/64 says:

    For this purpose, an 48-bit MAC address is expanded to 64
    bits as described in [RFC7042].

It wasn't entirely apparent to me what part of 7042 covers that. It
would be helpful to provide the section where this aspect is specified.
(After some study I guess that it is section 2.2.1.)

(7) MINOR: (Section A.2)

I believe that the values of both 'Length' and 'Address Sets End' are
too small by 7 - presumably because they forgot to count the fixed
fields. This also applies to the "alternative" using explict AFN encoding.

(8) NIT: (Section A.2)

Based on a very quick reading, ISTM that section 2.2.1 of 7042 suggests
that the IPv6 addresses being constructed this way should start with
0x02 rather than 0x20. But I'm far from sure I understand this correctly.

(9) NIT: (Section A.2)

There seems to be a typo in the following:

    The OUI would them be supplied
    by a second Fixed Address sub-sub-TLV proving the OUI.

I think "proving" should be "providing".