Re: [DNA] DNA meeting at IETF71

[JinHyeock Choi <jinchoe@gmail.com>]

Dear Bernard

Thanks for your kind feedback. Please find my in line comments.

> > I think there should be a way to indicate a DNA link.
>
> One question for simple DNA is whether we are attempting to determine
> a "link" or just connectivity to a particular router interface.  This is a
> subtle,
> but potentially important difference.

I agree. The difference is subtle but important.

> My impression had been that the goal
> was the latter.

Previously the first was the goal. DNA scheme was supposed to verify a
host's current IP configuration such as IP address by checking for
link change.

If that no longer holds and now the second is the goal, WG would
better make a clear and explicit declaration accordingly. Also we need
to discuss pro & con of two.

> > Otherwise the host has no way to determine whether to use
> > simple DNA or ordinary ND.
>
> Given that NUD uses unicast packets, would there be situations in
> which "simple DNA" is not supported on a link, but "ordinary ND" is?

Before discussing this in detail, we'd better make it clear how simple
DNA operates. However, allow me to express my impression based on the
current draft.

I assume that
a simple DNA host determines a link change
when it receives an RA with no known prefix from below.

   Additionally, where a host receives a solicited router advertisement
   after sending an RS for the purpose of detecting network attachment,
   and this prefix contains only prefixes which are disjoint from known
   advertised prefixes, the host SHOULD conclude with high probability
   that it has moved to a different link.

However, in non DNA link, it doesn't mean a link change to receive
an RA with no known prefix, so a host would better not to decide
a link change  based on prefix comparison.


> One of the design principles of "simple DNA" is that it is never slower than
> normal IPv6 address acquisition process (e.g. RS./RA + DHCPv6 if necessary).
> If that is true, then should not be a need to make this determination.
> Simple
> DNA is attempted, and if it fails, it fails.

I understand that DNA aims to improve not only speed but also precision.
When a host receives a solicited RA, it need to make a different decision
according to whether it's in DNA link or not.

For example, when the host receives an RA with no known prefix,
if it's in a DAN link, it can assume with high probability a link change.
However, if it's not in a DNA link, there is no guarantee of high probability.

Thanks for your kind consideration.

Best Regards

JinHyeock