[DNA] Prefix information for link identification in DNA

[Brett Pentland <brett.pentland@eng.monash.edu.au>]

Hi DNA group,

There doesn't seem to have been much discussion lately on how to
move forward on a solution document.

At the last meeting there was roughly equal interest in two
mechanisms for link identification.  Those mechanisms are fundamentally
similar because they are both based on identifying a link by the
address prefixes that are being advertised by the routers on the link.

Though it was in reference to CPL, Keith Moore raised an issue that is
relevant to both of the link-identification solution proposals (LinkID
and CompleteRA).  He thought that it is questionable to assume that:

    1) Every network has a router.
    2) you can name a network using one of its prefixes.

(I hope that's right Keith - I pulled it out of the minutes)

Both of the proposed solutions use prefix information for link
identification, though neither relies on using a single prefix as
the link name for eternity.  Both allow the prefix(es) used to
change over time and are designed so that hosts can cope with
those changes.  However both also rely on routers to distribute
this information to hosts which falls apart if assumption 1)
above is invalid.

So if a link has no router and no prefixes except the link-local
prefix (which will be the same for all links) we have a problem.

I'm not sure what we can reasonably do at layer 3 if there is(are)
no router(s) present on the link to help the hosts identify the
link.  (Any ideas?)

If there are routers present then I think that prefix information
should be usable for identification.  If there are no global
prefixes then maybe the solution can require that routers configure
a ULA prefix (it might simply be used in an identifier and not
advertised in a PIO)?

I guess I would like to find out:

1. Does anyone think we can come up with a reasonable solution at
    layer 3 that can deal with links that have no routers?

1a. Should we?

2. If we can't/won't, and thus we can make use of routers in
    identifying other links, can anyone explain why prefix
    information cannot/must not be used in some way for identifying
    links?  (Note that using prefix information doesn't require
    picking a single fixed prefix to act as the identifier)

2a. If it is *possible* to use prefix information in general to
     identify links, then is there a good reason why we should
     design an identification scheme that allows the use of non-
     prefix information for link identification?

Opinions?
Are these the wrong questions to ask?

Brett.