Goals for Network-Based Localized Mobility Management (NETLMM)
draft-ietf-netlmm-nohost-req-05

[Ballot comment]
I guess I'm less picky about 'goals' documents that some other ADs. My view is that this is a useful document as long as nobody tries to interpret it rigidly as 'requirements.'

Two editorial matters from Gen-ART review by John Loughney:

1) Contributor section at the beginning seems odd, and the second
sentence seems a bit gratitous.  I'd suggest moving it to the
acknowledgement section and dropping the last sentence.
     
    Gerardo Giaretta, Kent Leung, Katsutoshi Nishida, Phil Roberts, and
    Marco Liebsch all contributed major effort to this document. Their
    names are not included in the authors' section due to the RFC
    Editor's limit of 5 names.

2) The 1 sentence abstract is a bit weak, ... it could state a bit
more.

[Ballot comment]
I guess I'm less picky about 'goals' documents that some other ADs. My view is that this is a useful document as long as nobody tries to interpret it rigidly as 'requirements.'

[Ballot discuss]
I think the whole section on location privacy needs extensive work.  Some of the threats described therein have nothing to do with geography, and the assumptions on which the geographic threats are based are not clearly stated.

The GeoPRIV folks have worked in the area of geolocation privacy for some time, and they may well be able to provide help in how to clearly describe these issues.

[Ballot discuss]
I concur with the security directorate review's concern that this
document cannot be reviewed before draft-ietf-netlmm-threats is ready
to come to the IESG.  I'd like to see these documents come back on the
same agenda for concurrent review.  I may add additional comments to
this discuss at that time.


Section 2.0 claims that multicast is out of scope for the goals
discussion.  I don't think we can do that.  The implication of this
specification is that we're going to support local mobility management
for unmodified nodes.  That Other than a mechanism to signal a routing
update, these nodes will not change, but must continue to function.
That means that we must preserve the IP service model and these nodes'
ideas about their interfaces to the L2 network.  Part of that service
model is multicast.

Several of the goals talk about reducing signaling.  I think these
goals are misguided.  Reducing signaling in and of itself is nice to
have, but it seems like extensibility, security and modularity are
higher priorities than reducing signaling for its own sake.  However
the signaling today is presumably causing real problems--presumably
related to latency and other things.  Please actually describe these
problems as the goal and describe reducing signaling as the solution
or a derived goal.

Goal 2.6 seems inappropriate in two ways.  First, I don't think no new
security is a reasonable goal for its own sake; it may fall out of no
software changes and that's fine, but "we don't want to make things
more secure," doesn't seem like the message we're trying to send.
Another part of 2.6 is the idea that we don't want additional
credentials on the nodes because that would hurt deployment.  At one
level that seems like a reasonable goal.  I'm concerned about how
realistic it is in practice though for links that will require
authentication of the IP-level movement detection.  We're going to
need some credentials to make that work.  I don't understand how those
credentials would be any different than credentials required by
localized mobility management signaling.  Note that deploying this
security will become more important in the netlmm case than in the
current case because of the possibility of stolen IP addresses.  So,
in short, I think you have the security deployment problem no matter
how you look at it.

[Ballot comment]
My main concern with this document is that many of the goals are pretty
  vague and cannot be objectively verified. As such, I don't see what
  purpose this document really has. Many of the goals are generic
  and can apply to pretty much any protocol and there is significant overlap
  between different goals.

  The gap analysis in Section 8 would be useful. However, in its
  current form, it doesn't present either the different existing
  proposals or the arguments for why they evaluate in a certain
  way for the different goals in sufficient detail to understand
  the conclusion.


Section 2.1, paragraph 2:
>      A goal of the protocol is to reduce the loss of accurate forwarding
>      to reduce interruptions which may cause user-perceptible service
>      degradation for real time traffic such as voice.

  The previous paragraph states a stricter goal ("IP handover time
  should be between 40-70 ms") than this. "Reduce loss of accurate
  forwarding" is pretty vague.


Section 2.2, paragraph 4:
>      The goal is that handover signaling volume from the mobile node to
>      the network should be no more than what is needed for the mobile
>      node to perform secure IP level movement detection, in cases where
>      no link layer support exists.

  "No more than what is needed" is not an objective criteria.


Section 2.5, paragraph 1:
>      Therefore, any solutions for localized
>      mobility management should minimize signaling within the wired
>      network as well.

  There is overlap among goals 2 ("Reduction in Handover-related
  Signaling Volume"), 4 ("Efficient Use of Wireless Resources") and 5
  ("Limit the Signaling Overhead in the Network"). Because pretty much
  all signaling will be handover-related, goal 2 already subsumes goal 5
  and especially goal 4. Additionally, some of these goals ("minimize
  signaling", etc.) aren't verifiable


Section 2.7, paragraph 3:
>      The goal is that the localized mobility management protocol should
>      not use any wireless link specific information for basic routing
>      management, though it may be used for other purposes, such as
>      identifying a mobile node.

  Why is this goal specific to wireless links? I'd recommend to erase
  the word "wireless" in Section 2.7 (and possibly the entire document).


Section 2.10, paragraph 0:
>  2.10 Re-use of Existing Protocols Where Sensible (Goal #10)

  How is this specific to NETLMM? How is this objectively verifiable?
  Candidate for removal.


Section 7.0, paragraph 0:
>  7.0    Informative References

  No normative references?


Section 8.0, paragraph 0:
>  8.0  Appendix: Gap Analysis

  This appendix should probably be expanded and become a standalone
  document. In its current form, it doesn't present either the different
  existing proposals or the arguments for why they evaluate in a a
  certain way for the different goals in sufficient detail to understand
  the conclusion.