ietf
[Top] [All Lists]

Re: N:N multicast with extra address space?

2001-04-20 01:00:02

there's a discussion on how to make some simple classes of
assymetric multisender apps work with
SSM, but there's not really anything useful
for genuine multi-peer applications - what is needed is to
revitalise the work on bidir pim, and then retrofit the SSM
addressing (.e. what we proposed in rama/sm work) and maybe now is
the right time to do that since the SSM stuff is seeing deployment,
so we wont be distracting people from their main agenda anymore.

In message <3ADE68A8(_dot_)A93221EC(_at_)act(_dot_)cmis(_dot_)csiro(_dot_)au>, 
Hugh Fisher typed:


I'm building applications for collaborative 3D graphics
using SRM (actually LRMP) multicast on the local Ethernet.
They're peer to peer systems, not in the new buzzword of
the day sense, but in all being equal participants rather
than a client/server design.

My systems have multiple senders, but a per-group routing
tree rather than per-source is fine. In the future I'll
want (like many VR/distributed sim folk) to be able to
allocate from a large range of multicast addresses, say
12 bits or more. Waiting for IP6 isn't an option.

The "Interconnections" 2nd ed book describes a shared tree
multicast with 8 byte group IDs that would fit my needs
very nicely. Can anyone point me to a working group/mailing
list where I can find out more about this?

-- 
    Hugh Fisher
    ANU/CSIRO VE Lab


 cheers

   jon



<Prev in Thread] Current Thread [Next in Thread>