Paper review : IP Multicast Channels: EXPRESS Support for Large-scale Single-source Applications (HC99)
Reviewer : Hai Fang (email@example.com)
To provide an IP multicast model based on the conventional model and solve the problems on accounting,
security, and scalability, etc.
This paper specifies the single-source multicast as an extension of IP multicast, demostrates its
advantages for large-scale multicast applications, and adds the accounting support.
- Main ideas
- The EXPRESS channel model is a straightforward extension to the conventional IP multicast model,
and can be implemented using a simple protocol that requires no change to the current mechanisms
- The EXPRESS channel model has many advantages w.r.t. the old group model, e.g., it provides a
basis for charging, provides easier access control, and it's easier to scale.
- Significance rating: 3
This paper proposes several drawbacks of the traditional group model for IP multicast, and presents
the new channel model, which is based on the single-source multicast model. This is a promising try
on the modification of the IP multicast model, but it does not solve all the problems.
- Convincing rating
The authors compare the new model and the protocol with the existing ones; hence analysis the cost and
scalability with some calculations; they also simulate the counting overhead and the proactive counting.
Now that the authors claim they care about the scalabilty problem most, and examplify the sports-tv.net
several times in the paper, why don't they provide an experiment in the scale as large as the sports-tv.net
Although the channel model seems like better than the original IP multicast group model, I still doubt
the scalability of the EXPRESS multicast, especially for the case of multi-source applications. Hence,
I do not know whether it's easy to be deployed by the Internet in the near future.
How to provide a scalable multicast model is a challenging topic, all the other problems on multicast,
such as accounting, safety, efficiency, etc., are all facing the scalability problem.