Re: [DNA] RE: Review of draft-krishnan-dna-simple-01.txt

[JinHyeock Choi <jinchoe@gmail.com>]

Dear Suresh

> I have to admit that when we wrote the draft, we did not consider the
> precedence between the NS/NA tests and the RS/RA test. The way I see it
> today, an RA from a router containing disjoint prefixes will only
> override a positive NA confirmation from the same router and other than
> this there will be no interaction between the NS/NA tests and te RS/RA
> tests. I will give this a little more thought and come up with some text
> on this regard.

I'm not sure such an intricate deliberation is necessary. In the first
place, simple DNA is supposed to be simple even though it may not
cover some corner cases.

For example, simple DNA can work as below.

1. In a link, a host generates 1) router list and 2) prefix list.

2. Upon initiating DNA, the host performs 1) NUD with routers and 2)
Router/ Prefix discovery.

3. If a solicited NA from a known router arrives,
       the host assumes that
             1) it still remains at the same link and
             2) its addresses are still valid (but it may perform oDAD
in case).
       and turn-off DNA.

4. If no solicited NA arrives and an RA with no known prefix arrives,
       the host assumes that it has moves to another link and
       invalidates all its addresses.

Then we don't have to worry about RA sources but the above will work
for most of the cases.

Thanks for your kind consideration.

Best Regards

JinHyeock