Comments on draft-eronen-mobike-mopo-00.txt

Pasi,

I liked your draft! I think your approach of having
a set of separate features - independently usable even
in a non-mobile situation - is a good one. I also liked
the NAT prevention feature a lot! The path testing
approach you have is also good, I think.

A couple of comments:

o Section 1.2. If I am not mistaken, one of the features
   that you are also not doing is the movement of IPsec
   SAs one by one. You move them all at once, right?

o Section 7.3: I'm not sure I understand exactly
   what "when the window size allows" implies. Can we
   get into an eternal wait if there's something else
   going on at the same time as we move?

o Section 8.1: Couldn't you condense the first two bullets
   by saying that if the responder is not behind a NAT
   and is non-mobile, then the initiator can be multihomed,
   mobile and behind any type of a NAT? If this is the case,
   then I believe there's sufficient functionality.

o We need more details to what happens when you receive
   a NAT_PREVENTION payload and there has been an address
   change in transit.

Editorial:

o Section 6: s/not addresses/not all addresses/

--Jari