Stig Venaas, UNINETT


Multicast update, 6NET and more




6NET


Almost all NRNs and almost all other partners have native IPv6 multicast.


Experimenting with embedded-RP and scoped BSR
draft-ietf-mboned-embeddedrp-00.txt draft-ietf-pim-sm-bsr-03.txt


Embedded-RP available from Cisco in supported images, e.g. latest 12.3T.


Hoping for other vendors to follow.


I can provide patch for pim6sd, and *BSD included in some embedded-RP
tests. Want to try it on larger scale in M6Bone. Mail m6bone list or me
if interested.


Running beacons and doing some applications trials and demos using
multicast.




MLD Proxies


If you want multicast connectivity through a router that can't do PIM,
you may make it do MLD proxying instead.


This means doing MLD queries on downstream interfaces. On upstream
interfaces, act as host and send reports for all groups that are
reported on downstream interfaces. Forward multicast received on an
interface to all interfaces with listeners and to upstream interface
(if not received from there).


No IPv6 PIM on Linux yet (might be soon), in the meantime MLD proxy
useful.


Recently two Linux implementations have surfaced. Both work totally in
userspace, just a single process, and could easily be ported to other
systems.




1. By Jeroen Massar, see http://unfix.org/projects/ecmh/


ECMH used to provide multicast in http://www.sixxs.net/


A tunnelbroker service using Linux. Some PoPs now use ECMH to allow
people to receive multicast over their tunnels.


HEAnet (the Irish NRN) is involved, running ECMH on their SIXXS PoP.


Interested in connecting to M6Bone.



2. By Christophe Jelger <jelger@dpt-info.u-strasbg.fr> in 6NET


Christophe did some interesting test of multicast combined with MIPv6,
see http://www.6net.org/publications/deliverables/D4.1.5v1.pdf


MLD proxy used at Linux HA in his tests.


Multicast was forwarded over tunnel between HA and MN




Reliable multicast


http://www.atm.tut.fi/mad/ is an open-source implementation of
draft-ietf-rmt-flute-07.txt


Uses forward error correction to provide some reliability


Should maybe experiment with this


Can we find some useful scenarios?