Re: [multimob] PMIPv6 extension for multicast

[Hitoshi Asaeda <[e-mail]>]

Hi Behcet,

>> > We revised our PMIPv6 extension draft as follows:
>> > 
>> >  http://tools.ietf.org/html/draft-asaeda-multimob-pmip6-extension-06
>> > 
>> This document addresses the tunnel convergence problem by an  RPF
>> lookup algorithm. Although the approach is straightforward,  this
>> condition assumes both MAG and LMA act as PIM-SM routers.
> 
> In Section 3.2 you have:
> 
> The MAG selects only one upstream link for a multicast channel by
> the Reverse Path Forwarding (RPF) algorithm even if the MAG has
> established multiple  M-Tunnels to LMAs.  This does not cause the
> tunnel convergence problem, because duplicate packets are not
> forwarded to the MAG. This means that MAG does not send any new
> joins to the same group upstream, right?

Right.

> I agree that it would solve the problem but does it create new
> problems, like what happens if the member MN leaves the group and no
> more data can be received while other MNs possibly wanted to continue.

MAG maintains the membership state with MLD.
It is mentioned in section 6 and section 8.1.
For instance, section 8.1 says;

  "If p-MAG thinks that the moving mobile node is the last member of
   multicast channel(s), p-MAG confirms it by sending IGMP/MLD query.
   After the confirmation, p-MAG leaves the channel(s) by sending the
   PIM Prune message to its upstream router."

Well, IGMP is not relevant here, so I'll remove the word "IGMP" from
this sentence in the revised version, though.

>> The proposed  mechanism aims to be flexible and general; it could
>> cooperate with other fast  handover mechanism. As one of the examples,
>> this document refers the idea of  CXTP with multicast extension (*1)
>> and illustrates the scenario to cooperate  with it.
>> (*1) draft-vonhugo-multimob-cxtp-extension-00.txt
> 
> I am confused here.
> If we use draft-vonhugo-multimob-cxtp-extension then the handover in
> Section 8 would not be needed? Is the technique in
> draft-vonhugo-multimob-cxtp-extension stand alone and would achieve
> fast handover by itself or not?

Our draft proposes three possible ways for handover; 1) Policy Profile,
2) PBU-M/PBA-M, and 3) CXTP extension. Either one can be used.
And the above CXTP draft is a stand alone solution as it potentially
also works with other approaches.

Regards,
--
Hitoshi Asaeda