Paper Review: < CRSZ01 >>
Enabling Conferencing Applications on the Internet using an Overlay Multicast Architecture
Reviewer: Oleg Elkhunovich
While IP multicast has been proposed 10 years ago, it has yet to be widely deployed. There is a lot of concern over its support, congestion control and reliability.
Authors present some experimental evidence that an alternative framework to IP multicast - End System multicast can work very well. They use audio and video applications to explore the performance of this framework in the real world Internet.
It is important to adapt to both latency and bandwidth when constructing overlays for video conferencing.
End system multicast is a viable framework for enabling the performance of demanding audio and video applications.
The distinguishable characteristics of conferencing applications of performance requirements, gracefully degradable, session lengths, group characteristics and source transmission patterns
This paper does a good job advocating the use of the End System Multicast. Authors give some convincing evidence that this framework would work well for demanding applications.
The evaluation of the data is very thorough and experiments are pretty realistic.
A number of questions remain to be answered. One is whether the overhead of active measurements scales well to groups much larger than twenty members. Another question is how much time self-organizing protocols must take to converge on efficient overlays given no initial network information.