Early Review of draft-ietf-trill-smart-endnodes-02
I have been selected as the Routing Directorate QA reviewer for this
draft. For more information about the Routing Directorate, please see
At this stage, the intend of the following is not to discuss the WG's
decision about the I-D, but rather to help improving its content.
Please not that I am deeply familiar with TRILL, but this was an
interesting opportunity for me to browse a few RFCs from the WG and
learn about the technology. Bottom line: I may raise some rookie's
comments, feel free to educate me if appropriate.
The I-D is well focused on a protocol specification, addressing a
requirement with clear motivations. I must acknowledge a strong
improvement from 01 to 02, both in terms of readability and
specification. Beyond the list of minor issues below, I have one major
concerns with 02. All the same, from where I stand, it does not look
like a deep change and may not be that difficult to address.
As far as I understand, the I-D now defines a new protocol among those
to be carried within the RBridge Channel. This is consistent along
sections 4.1 (encapsulation) and 8 (IANA allocations). What is now
inconsistent is the use of IS-IS TLVs and sub-TLVs, probably because the
former version seemed more IS-IS based. Several related problems:
1- The TRILL GENINFO mentioned in the I-D is an IS-IS TLV, the way it
should be used in the Smart-Hello protocol is unclear.
2- The I-D is requesting a new APPsub-TLV from the GENINFO above, i.e.,
within the IS-IS registry. (a) A new protocol being defined, if the
IS-IS registry is to be used, this needs to be clarified. (b) It is not
clear either why the GENINFO top TLV is still useful since the
information is not conveyed using IS-IS.
3- The I-D specification reuses yet another IS-IS TLV (242) to be
included in "Smart-Hellos". While defining a TLV content aligned on an
existing IS-IS TLV seems reasonable (the PCE WG faces a same situation
with PCEP sometimes pointing to RSVP registry), this TLV must have its
codepoint allocated as part of the "Smart-Hello" protocol, and values do
not have to match.
4- Note also that IS-IS registry associates TLV with messages types
(i.e., IIH, LSP, SNP, etc), that IIH does not support TLV 242 and that
Smart-Hellos are not part of the list. This is likely to be solved as
part of addressing 3-, once that TLV is allocated in a Smart-Hello registry.
- The I-D is currently ST. For a protocol specification, there are too
few occurrences of RFC 2119 keywords. Either the I-D is not really ST
and the field should be updated, or the keywords, as well as RFC 2119
reference (which got dropped from 01 to 02) and introductory quote,
should be included.
- Capitalized 1st letters on "Smart Endnode" and "Smart-Hello" have been
generalized, just a few ones were missed (e.g., in sections 2, 4.1,
etc.). The choice is less clear on the "e/Edge RBridge" phrase: please
make it consistent along the I-D.
- Some character spacing should be fixed.
- All authors' first names have a trailing period to be dropped.
- The section extensively use some of the acronyms only defined later.
The terminology section should be moved before it.
- s/solution proposed/solution defined/
- s/the below figure/the figure below/
- s/attached RB/attached RBridge/
- s/if there was no endnode entry/in the same situation/
- Figure 1: acronyms are enough, full phrases useless and heavy.
- s/RB 1/RB1/ (on figure 1)
- s/issues a Smart-Hello/issues a message called "Smart-Hello"/
- required acronym expansions (fine since terminology if moved before):
"LSP" (the Routing area loves RFC 5513!), "E-L1FS FS LSP"
- To be moved and extended (see above and below)
- The "TRILL switch" phrase appears in various definitions: is it still
needed beyond defining "RBridge"? I imagine it was required in early
days of TRILL work, but now that has its own terminology, sticking to
"[edge/transit] RBridge" looks clearer. Obviously, this should be
enforced along the I-D, i.e., avoid "TRILL switch" in the remainder of
- s/smart end node/Smart Endnode/
- s/both for Smart-Hellos sent by Smart Endnodes and for Smart-Hellos
sent by Edge RBridges/for Smart-Hellos sent by both Smart Endnodes and
- s/an 8-bit size and type field/an 8-bit size and 8-bit type fields./
- As pointed out as "main issue", RFC 6823 does not seem applicable here
(though the idea may).
- I am rather used to 32-bit-aligned TLV figures, but the display format
used here does not harm.
- I wonder why not choosing a fixed header format, using the larger
size, i.e., 6-bit "RESV" and 24-bit "VLAN/FGL Data Label" in all cases.
I see more trouble in handling 2 different sizes rather than carrying
padded values. Moreover, the 2-bit RESV option may quickly become too
small, since using one of these bits is already considered in section 6.
- s/IS/node/ (Note that if the IS acronym was really to be used, it
should join the terminology section.)
- RFC 2119 keywords to be considered all along.
- s/D either queries/SE1 either queries/ (if I understood well)
- s/DRB/designated RBridge/ (used once, no need for the acronym)
- In section 5.1, the very last sentence of the last paragraph should
become penultimate: easier to read all expected behaviors first, and the
crash/restart case only then.
- The phrase "port with a [S]mart [N]ode", used a couple of times, read
weird. "Port connected to a Smart Node" would ease parsing.
- s/MAC (or Data Label)/MAC or Data Label/
- "would be dropped": an example of loose wording, RFC 2119 keywords are
- s/TRILL encapsulation/TRILL packet/
- "Normal endnodes" is not defined (implying smart ones are abnormal?),
"non-smart endnodes" should be preferred and used each time.
- s/decapsulate the frame/decapsulate the TRILL-encapsulated frame/
- s/two copies/two formats/
- s/A Smart Ennodes/a Smart Endnode/
- s/A normal (non-smart) endnode/A non-smart endnode/
- s/RECOMMENDED/recommended/ (That is deployment guideline, not protocol
- "Another solution is..." Again, does not fit ST.
- Supposing that the I-D is actually on ST, not less than 3 possible
solutions are listed! If there is relevant work in progress, then point
to it and stop. If a mechanism is really missing, then define only one.
- s/RBridge RB1 and RB2/RBridges RB1 and RB2/
- s/multi- homing/multi-homing/
- Including a value is usually a bad idea at this stage. Please use
"TBD" or start an early allocation procedure to have a stable value to
- As mentioned above:
* Is it really an APPsub-TLV from IS-IS GENINFO that is needed?
* IS-IS TLV242-like should be allocated as part of Smart-Hello.
- Are those references all normative? I feel I have managed to
understand the I-D without diving into all of them. (And beware of
Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.
This message and its attachments may contain confidential or privileged information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.