All things related to multicast

[Hitoshi Asaeda <>]

Hi Marc and all,

Thanks for referring the following I-D.

> http://www.sfc.wide.ad.jp/~asaeda/paper/draft-liu-multimob-igmp-mld-mobility-req-00.txt

I also provided a mobility extension draft, for which I've been
working for its revision and will resubmit it within a short period.

http://www.sfc.wide.ad.jp/~asaeda/paper/draft-asaeda-multimob-igmp-mld-mobility-extensions-00.txt

Anyway, SSM is necessary and indispensable for the current and
future multicast communication.
IGMPv3/MLDv2 are the "required" implementations to enable SSM.

One drawback of supporting IGMPv3/MLDv2 is its implementation
cost. All hosts need to implement many functions and complex state
transition caused by source-filtering functions with INCLUDE and
EXCLUDE mode operations, where the source-specific EXCLUDE mode
operation is not usually used.

As we proposed in the another draft;
http://www.sfc.wide.ad.jp/~asaeda/paper/draft-ietf-mboned-lightweight-igmpv3-mldv2-03.txt
Lightweight-IGMPv3 (LW-IGMPv3) and Lightweight-MLDv2 (LW-MLDv2) are
fairly reasonable for supporting SSM, because they simplified
functions by eliminating the source-specific EXCLUDE mode operation
and hence it's kernel implementation is very simplified.

How the host-side kernel implementation is refined?
Well, you may want to check both IGMPv3 and LW-IGMPv3 implementations;
http://www.sfc.wide.ad.jp/~asaeda/igmpv3/
http://www.sfc.wide.ad.jp/~asaeda/LW-IGMPv3/

The last note; if Apple decides to support LW-IGMPv3/LW-MLDv2, I'm
glad to contribute these implementations.

Thanks,
--
Hitoshi Asaeda