Intermediate System to Intermediate System (IS-IS) Extensions for Advertising Router Information
draft-ietf-isis-caps-07

IANA Last Call Comments:

**  IANA has a question about the action required upon
approval of this document.  See below. ***

------
As stated in the IANA Considerations section of this
document, upoon approval of this document IANA will
complete a single action.

In the IS-IS TLV Codepoint registry located at:

http://www.iana.org/assignments/isis-tlv-codepoints

In the TLV Codepoints Registry, IANA will assign a new
IS-IS TLV code-point for the newly defined IS-IS TLV
type named the IS-IS Router CAPABILITY TLV.

Name Value IIH LSP SNP Status/Reference
---------------------- ----- --- --- --- ----------------
IS-IS Router CAPABILITY TLV tbd ? ? ?

IANA needs to know what the entries should be for the
IIH, LSP, and SNP columns for the new IS-IS Router
CAPABILITY TLV.

[Ballot discuss]
In response to the SecDir Review from Joe Salowey, the authors have
  agreed to the following Security Considerations text written by Joe:
  >
  > Any new security issues raised by the procedures in this document
  > depend upon the opportunity for LSPs to be snooped and modified,
  > the ease/difficulty of which has not been altered. As the LSPs may
  > now contain additional information regarding router capabilities,
  > this new information would also become available to an attacker.
  > Specifications based on this mechanism need to describe the security
  > considerations around the disclosure and modification of their
  > information. Note that an integrity mechanism, such as one defined
  > in RFC3567 or draft-ietf-isis-hmac-sha, should be applied if there
  > is high risk resulting from modification of capability information.
  >
  I suspect that an RFC Editor note is the easiest way to proceed.

[Ballot comment]
Section 1., paragraph 5:
>    The use of IS-IS for Path Computation Element (PCE) discovery may
>    also be considered and will be discussed in the PCE WG.

  RFCs shouldn't typically discuss what WGs at the time did or didn't
  do. Remove paragraph.

[Ballot comment]
Header line:
Proposed status: Standard
should be
Intended status: Standards Track

Editorial points from Gen-Art review by Francis Dupont:

- 1 page 2: I know what is IS-IS but a random reader likely doesn't:
  add a reference to the ISO IS (i.e., [IS-IS]) in the first sentence.
  (and perhaps add IS-IS-IP too, cf last comment).
- 1 page 2: "two" examples numbered from 1 to 3 (:-)!
- 2 page 3: I don't understand the "to prevent TLV looping" but
  perhaps it is not "TLV" looping?
- 3 page 3: I don't like the term "domain flooding scope", it seems a
  bit redundant to me.
- 3 page 4: "stale capabilities information A system"
                                            ^^ ", a "?
- 9.1 page 6: is ISIS-TE really a normative reference? BTW there is no
  reference in the text (same for the previous item, there is a nice
  checker for this: I suggest to use it...)

IANA Comments:

ANA has a question about this document. When the entry is
to be made into the IS-IS TLV Codepoints registry located at:

http://www.iana.org/assignments/isis-tlv-codepoints

Name Value IIH LSP SNP Status/Reference
---------------------- ----- --- --- --- ----------------
IS-IS Router CAPABILITY TLV tbd n y n [RFC-ietf-isis-caps]

IANA Last Call Comment:

IANA has a question about this document. When the entry is
to be made into the IS-IS TLV Codepoints registry located at:

http://www.iana.org/assignments/isis-tlv-codepoints

What are the values to be entered for the "IIH," "LSP, and
"SNP" entries?

Name Value IIH LSP SNP Status/Reference
---------------------- ----- --- --- --- ----------------
IS-IS Router CAPABILITY TLV tbd ? ? ? [RFC-ietf-isis-caps]

IANA understands that, once this question is answered, this
will be the only action required upon publication of the
document.

AD comments, sent on 19-Jul-2005:

I'm quite happy with this draft--looked through the WG discussion on the
seq numbers (which was my main question) and saw that it was discussed in
detail, so we're good. Minor comments I have were

> 3. Elements of procedure
>   
>    A router which generates a capability TLV MUST also generate a
>    Traffic Engineering Router ID TLV (134) at each level for which it
>    generates a router capability TLV.
>
>    When advertising capabilities with different flooding scopes, a
>    router MUST originate a minimum of two Router CAPABILITY TLVs, each
>    TLV carrying the set of sub-TLVs with the same flooding scope. For
>    instance, if a router advertises two sets of capabilities C1 and C2
>    with an area/level scope and routing domain scope respectively, C1
>    and C2 being specified by their respective sub-TLV(s), the router
>    MUST originate two Router CAPABILITY TLVs:
>   
>      - One Router CAPABILITY TLV with the S flag cleared, carrying the
>      sub-TLV(s) relative to C1. This Router CAPABILITY TLV MUST NOT be
>      leaked into another level.
>   
>      - One Router CAPABILITY TLV with the S flag set, carrying the sub-
>      TLV(s) relative to C2. This Router CAPABILITY TLV MUST be leaked
>      into other IS-IS levels. When the TLV is leaked from level-2 to
>      level-1, the D bit MUST be set in the level-1 LSP advertisement.

Ed: starting with "For instance..." in the second para and down here, the text
seems more of an example, and the 2119 language in it echoes already
defined procedures. If that's the case, it can be changed to "will" and
"will not" to avoid reader confusion.

>    When leaking Capability TLVs downward from Level-2 into Level-1, if
>    the originator of the TLV is a Level-1 router in another area, it is
>    possible that multiple copies of the same TLV may be received from
>    multiple L2 routers in the originating area. To prevent a router from
>    leaking multiple copies of the same TLV, the router performing the
>    downward leaking MUST check for such duplication by comparing the
>    contents of the TLVs.

for comparison, do you want to mask out the D bit, as an L1/L2 router can
receive the same TLV from the originator and another L1/L2?

And, of course, what's the implementation status for this? :)