Paper Review:
IP Multicast Channels: EXPRESS Support for Large-scale Single-source Applications
Reviewer: Robert Dugas
Problem
This paper addresses the problem of extending current IP multicast to become
more scalable, accountable, and secure.
Contribution
The author proposes an EXplicitly REquested Single-Source multicast system
which includes adaptations to address the problems mentioned above
Main Ideas
Multicast content should be divided into channels, delivery services designated by the sender and the channel destination address
Multicast recipients would subscribe to particular channels
Subscription protocol would account provide access control as well as
user counting
Critique
Significance:4
Although not supported by extensive testing the multicast protocol
proposed here could potentially become a widely used web content distribution
and billing model
Methodology:
Although some simultion results were included at toward the end of the paper,
the paper is largely theoretical, with algorithmic bounds serving as performance indicators.
Limitation:
As mentioned above, this paper lacks extensive real-world testing to show
that the proposed protocol extensions would work on a large scale. In addition
the cost calculations performed are somewhat sketchy such as the 18 cent per
subscriber per year result.
Lessons:
The real lesson of this paper is that the current IP multicast model is not sufficient to support commercial, broadly distributed content delivery. The model proposed is a step in that direction.