Re: Simple Multicast - building a case for a BOF or WG

[Radia Perlman - Boston Center for Networking <Radia.Perlman@East.Sun.COM>]

Re; Ken's question:
>>will there be an attempt to merge SM and Express,
>>or simply have separate protocols for each.

I wouldn't want this perceived as an SM vs Express. You could think
of them as being mostly equivalent, which is an 8 byte ID. The
difference is whether you want to support bidirectional trees or not.
Politically it would be nice to consider it a merged group, whatever
the outcome of that decision (bidirectional tree support). I think
the spirit of the two are similar enough that people that like one
would probably like either. I'd think it would be less confusing to
the world, and more likely for this approach to succeed, if we didn't
create two separate, and worse yet, competing things.

So something that should be explored is how important multisender
groups are, and how much it costs to do them. You can do multisender
with unidirectional trees,
by building multiple trees (one per source) or unicasting
to the Root, which creates a hotspot at the root and wasted bandwidth for the
sender-Root path. You can also do single sender with bidirectional
trees, by specifying
that the policy for that group is that only the Root is allowed to transmit.

Dave Oran had mentioned something interesting, which was that instead of
a single protocol (or perhaps in addition to a single protocol), the WG
come out with documents which are "this is how to modify protocol
X to support this
mode". So there could be one for PIM-SM, for CBT, for BGMP. So what would
people think of producing "here's the delta from this protocol", rather than
a self-contained protocol spec?

Radia