Re: Laugh Test: Next Steps in Mobility BOF [nsiim] Request

[Thomas Narten <narten@us.ibm.com>]

Hi James.

> Below, a few comments for fine tuning. Let me preface my comments by
> saying I think this BOF is an excellent idea and I fully support
> it. Pekka Nikander and I submitted a draft to MIP related to this
> topic that might be appropriate for consideration by the BOF. I'll
> bring it up on the mipcharter list when it is set up.

OK

> > It is now a good time to look at the activities of the working group
> > and how to better structure future efforts.  Upon doing so, it is
> > evident that beyond their superficial resemblance, mobility for IPv4
> > and IPv6 differ considerably.  They are very different with respect to
> > their underlying technologies,

> I disagree with this as a blanket statement. On a global level, the
> basic architectural mechanism for handling global routing changes
> hasn't changed because it is dictated by the Internet routing
> architecture, and that hasn't fundamentally changed between IPv4 and
> IPv6. The global mechanism is a routing proxy to handle the global
> node identifier to routing identifier mapping. MIPv6 has filled in
> some optimizations, like route optimization (but there was even an
> (insecure) prototype for that in MIPv4), and done a much better job
> at specifying security and separating out AAA from the basic
> mobility mechanism, but the basic design hasn't really changed on a
> global level.

I'm not so sure how important it is to agree or disagree on how
different they are. MIPv6 has route optmization with return
routability. RR is new stuff. AFAIK, there really isn't interest in
MIPv4 for even bothering with route optimization anymore (though in
theory, it could pick up the basic RR design from MIPv6).

What I think is worth noting is that MIPv6 vs MIPv4 involves different
consituancies, and different immediate short-term issues. 

> What has changed, and rather drastically I think, is how MIPv6
> supports local link configuration and change, with the demise of the
> Foreign Agent.

I'm not sure I understand that. Link configuration is not a
MIP-specific issue and MIP just uses what's available.

> I think it is very important to keep the distinction in mind, and,
> in fact, lack of attention to this distinction has resulted in some
> proposals for inappropriately mixing the global and local signaling
> for MIPv6 extensions, IMHO. The local part is the most
> underspecified in the current design.

Well, there might be a number of MIPv6 extension proposals, but the WG
has been in the mode of keeping them all on hold until MIPv6 was
done. Part of this BOF is to look at what we will do with all the pent
up demand for these extensions.  I wouldn't assume any of them are
going to necessarily go forward in their current form.

> >This leads to a two-pronged approach
> > in which the basic idea is to divide the current working group into two
> > new ones: one concentrating on IPv4 mobility and the other on IPv6
> > mobility.
> >

> I'm not so sure I would make the cut in this way. I agree on the
> need to separate out IPv4, but I think a finer granularity cut in
> IPv6 might facilitate more modularity in the design. The draft I
> mentioned above makes one proposal for such a finer cut, there are
> others. An advantage of separating out into smaller, more focussed
> WGs aligned along the natural architectural bounderies in MIPv6 is a
> possible potential for achieving closure more quickly. Mitigating
> against this is the more experimental nature of some of the
> technologies, which should require proof of concept and measurements
> to verify validity before being standardized, and that takes longer.

FWIW, another idea that was tossed around was separating along the
lines of "deployment issues" vs. "protocol work". But that tends to
end up having the v4 vs v6 split without actually calling it out.

> > This purpose of this BOF is to discuss this proposal and to seek IETF
> > input on how each of the resultant working groups should be chartered.
> >
> > Another issue to be discussed is to figure out how to best determine
> > when items should be adopted by the working group(s). As an example,
> > when items are more of a research nature, it may be best not to
> > pursue them in the IETF as other organizations may be better suited
> > for this. Other potential criteria to apply are the existence
> > of independent implementations and/or deployment plans as  a measure
> > of interest in a given proposal.
> >

> This is an excellent topic but I'm not sure whether this WG is the
> right place for this discussion, as it is a topic that is applicable
> to any item that comes before IETF. The problem has been somewhat
> more severe in MIP, but I can't really see any possibility for
> resolution at this BOF and so it sets off my rathole
> detector. Moving this discussion to the Problem Statement WG might
> be a better bet.

I agree this is an issue for problem-statement as well. But we also
have an immediate problem of how to recharter the MIP work. I don't
think we can really wait on problem-statement to figure this all out
(if indeed it can). Part of the reason for making this a BOF is to
allow folk from outside the traditional MIP community to come in and
provide input. Input on this topic would be welcome.

> > The chairs also seek feedback on how best to interface with other
> > working groups, both within the IETF (e.g., to tackle ND optimizations
> > with IPv6) and elsewhere (e.g., to coordinate work on movement detection
> > with the IEEE).
> >
> > Chairs:
> >      Phil Roberts <proberts@megisto.com>
> >      Basavaraj Patil <basavaraj.patil@nokia.com>
> >      Gabriel Montenegro <gab@sun.com>
> >

> Actually, I would suggest getting someone other than the MIP chairs
> to do this.  Not that I have any problem with their performance as
> MIP chairs (they are all good people and have done an excellent
> job), but I think it might be useful to have someone else as BOF
> chair to avoid any perception of a vested interest in the
> outcome. Considering the wide breath that the topic of IPv6 mobility
> covers, I think it might even be appropriate to have an AD be the
> BOF chair. Perhaps Erik Nordmark?

Both ADs will be in the room. And FWIW, I'm not so clear that the
chairs know exactly what they want the outcome to be. Well actually,
one thing that has been made clear is that they inherited a WG and
culture that they feel, in retrospect, had a number of problems. They'd like
the recharter to also act as a reset on some of those tendencies. 

Thomas