Summary: CS740: Review of A Case for End System Multicast
Shijin Kong (CS login: krobin)
This paper aims to solve some long-term problems generated from IP multicasting. The major thought
of this solution is to move all multicasting functionality from networking devices into end systems.
This paper studies with Narada protocol, in which end systems self-organize into an overlay structure.
Both simulation and Internet experiments are applied to verify the optimization of overlay efficiency
by dynamically adapting. The potential benefits of transferring multicast functionality from end
systems to routers significantly are believed to be worth the performance penalty incurred.
While end system multicast has many advantages, there are still several issues to be resolved for
practical applications. Overlay approach to multicasting is efficient, but they will generate some
unnecessary traffic by sending same traffic onto same physical links. Another problem is that since
end system multicast is based on individual host, the latency for some hosts may be much greater than
the average latency in IP multicasting. Although experiments and simulations results shows end
system multicast achieves good performance for small and medium sized groups, performance of end
system multicast architecture scale to support much larger group sizes is still remained as an open
For today's network, how well end system multicast can be applied need to be evaluated. Both IP
multicast and end system multicast have their advantages and disadvantages. The original purpose of
multicast is to improve the performance of continuous unicast from single host. IP multicast provides
latency close to unicast while end system multicast resolve the management and overhead caused by