[DNA-BOF] review of draft-park-dna-ipv6dadopt-requirement-01.txt

[Jari Arkko <jari.arkko@kolumbus.fi>]

I read this draft on the way to Minneapolis and have
now found time to write an e-mail about the questions
and comments I have on it:

Semi-substantial:

- Move the begining of Section 4 (all the way up to
   the requirements) part so that the possible solution
   ideas come after the requirements.

- Why isn't there a requirement that puts a number
   on the speed that DAD must normally succeed (as
   far as protocol limits go). I would suggest this:

    RQ 0: The optimistic DAD procedure MUST NOT introduce
          more than 10 ms latency in the normal case.

- RQ 1 discussion is mostly irrelevant to RQ 1. Delete
   everything and put this in instead: "The optimized
   DAD mechanism MUST NOT break the RFC 2462 DAD mechanism,
   if used on the link by other nodes."

   It may also be an idea to think about RQ1-4 again
   and see if we can find a shorter or more fundamental
   requirement relating to the backwards compatibility with
   regular DAD.

- I disagree about RQ 8 -- its not an optiDAD problem.

Editorial:

- Abstract is lengthy and hard to read. Rewrite it as follows:

   "In order to generate a unique address, IPv6 nodes perform
    the Duplicate Address Detection (DAD) procedure. Payload
    packets can not be sent or received before this procedure
    is complete. In an environment where frequent movements are
    expected, this delay can be problematic. As the reduction
    or removal of the delay would be useful, an optimized
    version of DAD has been proposed. This document discusses
    the requirements for such "Optimized DAD" protocol."

- Section 1, s/In the current specifications, an .../An .../

- I would delete the second last paragraph of Section 1.

- I would also delete the second defined term from Section 2.

- Section 2, last paragraph... hmm... just import the defs
   here if they are used. Shouldn't be too many, I think.

- Section 3, reaplce "Note that this ... latency except other latencies ...
   such as ..." with

   "This document only considers the latency introduced by DAD.
    There are also other protocols that introduce additional
    network attachment latency, such as ..."

- Section 3, s/in a separate document as well/elsewhere/

- Rename PS 1 to "Introduction of Random Start Delays"

- Rename PS 2 to "Introduction of Delays through Response Windows"

- Section 3, the paragraph after PS 1 line. Replace "So, " with
   "According to RFC 2462, ". Also, delete the "If a node abides"
   sentence altogether, its repetition and seems to suggest RFCs
   not be followed.

- Section 4, remove the first two sentences of the second paragraph.
   Repetition. Its getting obvious by now that you don't like delays,
   but I think we all agree and you have made your point earlier ;-)

- RQ 3, s/NDP (Neighbor Discovery Protocol)/ND (Neighbor Discovery)/

- RQ 4, delete "not interfere ... a router has to".

- Can you provide some more information on what [RANDOM] is? Can't
   see if it is an I-D or what it is.

--Jari