[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
AW: RE : [DNA] Current IETF-Draft version - POLIMAND
Dear Greg,
dear all.
Thank you very much for your comments regarding the draft. We hope that
we can support the current work items in the DNA WG with our draft
submission and we are grateful for the ongoing discussion concerning L2/L3
interaction.
However, in this draft it is described how the interaction of L2/L3 can be
used to establish handovers based on a policy. We will think about to extend
this draft in a way that is more independent of MIP. Proactive handover
decision can be appropiate to force Neighbor Discovery/Neighbor
Unreachability Detection
for DNA, for instance. However, we think that it might be important to
inform the MN
on the fastest way that the MN will change its point of attachment. The
first
implementation of POLIMAND has shown that lossless handovers can be provided
by using
L2 information for vertical handovers in overlay networks.
Kind regards,
Stefan
> -----Ursprüngliche Nachricht-----
> Von: owner-dna@ecselists.eng.monash.edu.au
> [mailto:owner-dna@ecselists.eng.monash.edu.au] Im Auftrag von
> Gregory Daley
> Gesendet: Montag, 1. März 2004 16:19
> An: NJEDJOU Eric FTRD/DMR/REN
> Cc: Pekka Nikander; Stefan Aust; basavaraj.patil@nokia.com;
> Carmelita Görg; Niko A. Fikouras; Cornel Pampu; Dna; Greg
> Daley; mip6@ietf.org
> Betreff: Re: RE : [DNA] Current IETF-Draft version - POLIMAND
>
>
> Hi,
>
> I know that this has been cross posted, so I'll
> respond to both lists, but perhaps we can talk about
> specific issues in the individual mailing lists
> subsequently. I'd suggest DNA, unless the MIP6
> chairs wish to respond.
>
> If we need to discuss needs more generally for
> multiple interface configuration procedures or
> vertical handoff, it may be useful to take it up
> with the chairs of either or both WGs (depending
> on the circumstances).
>
> Please see my take below.
> ----- Original Message -----
> From: NJEDJOU Eric FTRD/DMR/REN <eric.njedjou@francetelecom.com>
> Date: Monday, March 1, 2004 9:51 pm
> Subject: RE : [DNA] Current IETF-Draft version - POLIMAND
>
> > Pekka,
> > The problem with MIP6 movement detection handover optimizations
> > (see http://www.ietf.ord/I-D/draft-daniel-mip6-optimized-vertical-
> > handover-00.txt for another draft) is that it is being considered
> > out of scope for DNA because of the mobile IPv6 issue raised and
> > out of scope for MIP6 because of the link triggers that has to be
> > taken into account. Actually, it is within the scope of both as
>
> I'd guess while this is frustrating, it's largely a statement
> of fact.
>
> Please inform me of any incorrect statements I make, since
> there may be subtlties I miss in the drafts.
>
> While the drafts being discussed are principally interested
> in handling multiple connectivity scenarios associated with
> wireless networks, neither MIPv6 (regarding Link hints) or
> DNA (Discussing global signalling) are currently chartered to
> work on these topics.
>
> It is indeed possible that the combination of the techniques
> is seen as the stumbling block by both groups.
>
> For example, if indeed it is necessary to provide vertical
> handover support for MIPv6, is it strictly necessary to talk
> about link layer interactions (polimand may talk about policy
> instead, Of course, it is always hard to standardize policy).
>
> Whether a document looking at multiple interface connectivity
> for MIPv6 would be useful or not (or whether there's much
> difference except the movement detection algorithm) I don't know.
>
> Particularly for DNA, we can't really discuss handovers,
> considering that this term principally refers to session
> restoration or continuation using mobility protocols.
>
> DNA is looking basically at signalling using mechanisms
> within the local link only. Any reference to other
> protocols (except in a confined way) can make documents
> look like they're (for example) MIP specific.
>
> If there's a difference in the configuration detection
> procedures when a host has multiple interfaces, that may be different.
>
> For example, a host may wish to wait passively for router
> advertisements on the GPRS interface (testing reachability
> when the host switches connections to that interface), and
> probes fairly aggressively on a WLAN interface.
>
> Any decision or action based on this the configuration
> detection is starting to push into uncomfortable areas as far
> as DNA's tight charter is concerned. Perhaps this is
> analogous to the reasoning used in MIP6?
>
> > DNA -link triggers/hints that will be catalogued/standardized
> > within DNA will also be used for Mobile IPv6 optimised movement
> > detection. Other ones may be needed for this specific purpose.
> > Better detecting network attchment/detachment is a requirement for
> > Mobile IPv6 better performance movement detection and handover.
> > These issues are therefore strongly linked.
>
> I think we have to be particularly careful about the word
> standardized in the previous paragraph. There's already been
> some out-of-band feedback to DNA that we should avoid delving
> into standardization of L2 oriented mechanisms.
>
> > MIP6 -Optimized movement detection an handover schemes are
> > essenttial for base Mobile IPv6 Operation, which means, the
> > protocol operation needs exploiting L2 triggers/hints. And MIPv6
> > implementors use L2 triggers/hints in a confusing way, if
> not at all.
>
> While the usage of hints may be unavoidable in some DNA schemes,
> there will still be some systems where it is not possible to
> the the kind of hints you desire. It would be short-sighted to
> design systems which only work if information from the
> link-layer is reliable and available.
>
> > We could be ping-ponging betwwen these two groups for years.
>
> I really hope not, but MOBOPTS isn't necessarily a place for
> killing ideas off to (no matter what some people say!).
> Our research team has (voluntarily!) submitted some ideas
> there for further discussion recently and gotten useful feedback.
>
> I don't even mind if people winnow their existing work down
> to a problem statement which is DNA centric. It may be that
> there's a chance that we'd think it out-of-scope anyway, but
> that's because the WG scope is very narrow.
>
> If the work is clear input into one of the documents on the
> DNA charter, that may change things.
>
> We're only likely to produce 4 documents this year
> (optimistically). so focussing on how the vertical
> handover/polimand work can directly help those is more
> immediately useful.
>
> As Pekka said, this may mean leaving MIPv6 specific parts to
> MIP6WG or MOBOPTS.
>
> Greg
>