Re: [DNA] Best current practice for DNA

[Pekka Nikander <pekka.nikander@nomadiclab.com>]

> For BCP for DNA, IMO, we should make a guideline for DNA
> just by using the existing standards, such as RFC2461 and MIPv6-24.

Exactly.  That is the idea.  No new mechanisms, just clarifications
on how to use existing ones most effectively.  Getting all the relevant
information into one place instead of being scattered over multiple
RFCs.

In addition to existing RFCs, we can also consider mechanisms that
are in the RFC editor queue or known soon to enter the queue, i.e.
before the DNA BCP will enter.  IMHO, we probably can also take
some material from rfc2461bis and rfc2462bis, but we should not
(necessarily) rely on them to get published before the DNA BCP.

Greg has identified a potential initial author for the draft,
but he will need help.  Hence, our current idea is to form a
group of people that will submit material to the initial author
and help him to write the -00 version of the draft.

> http://www.thinkonweb.com/sait/MD.pdf

Based on a very quick look, this looks like a very good input
to the BCP.  One thing that needs to be done, though, is to
clear the document from mobility terminology.  Remember, we
are looking at the situation just from the performance point
of view, considering all nodes, not only mobile nodes.

Focus is more on how to reconfigure a node quickly when
it come to a network whose configuration it already knows
than on how to create a new configuration quickly when a node
attaches to a new L3 link where it hasn't been before (and
whose configuration it doesn't know by other means).
Much of the focus is to ensure that no harm is done by
making (perhaps invalid) assumptions.

> Although it is not enough, it can be a stating point material
> for discussion, I think.

Yes.

--Pekka