From: Erika Rice (erice@cs.washington.edu)
Date: Tue Nov 09 2004 - 23:58:22 PST
"The PIM Architecture for Wide-Area Multicast Routing" by Stephen
Deering, Deborah Estrin, Dino Farinacci, Van Jacobson, Ching-Gung Liu,
and Liming Wei:
This paper presents an architecture, PIM (Protocol Independent
Multicast) for doing multicast when the group members are distributed
across the network in a sparse manner. The architecture has the
advantages that it supports different types of distribution trees,
retains the property of having receiver initiated membership, does not
depend on any one routing protocol, and uses soft state instead of hard
state so that it can adapt.
However, the discussion of the architecture and of all the values that
the routers need to keep track of highlights an inherent difficulty in
implementing multicast. Multicast must necessarily be widely supported
by routers. While some might object to this as a violation of the
end-to-end argument, those who really understand that argument can see
that multicast can only be implemented within the network; there is not
where the problem lies. The problem is exactly the practical one of
deployment. No matter how great the multicast architecture, deploying
it to the whole network will be difficult.
Another thing I find interesting is the lack of discussion of the model
of how receivers attach themselves to the multicast. They ask to join
and then they are part of the group. This prompts one to wonder if more
secure access could be added onto this scheme and how difficult such an
addition would be.
This archive was generated by hypermail 2.1.6 : Tue Nov 09 2004 - 23:58:23 PST