Paper review : Intergrated Services in the Internet Architecture: an Overview (BCS94)

• Goal
  To provide service for wide variety of applications, dealing with real-time traffic.

• Contribution
  This paper proposes a new architecture to encompass integrated service: the IS model which extends the original best-effort model and a reference implementation framework to realize this model.

• Main ideas
  • The old service model for Internet cannot manage resources efficiently at IP level, whereas the IS model may provide different kinds of service for different applications.

  • The framework includes four components: the packet scheduler, the admission control routine, the classifier, and the reservation setup protocol; the coodination of these conponents provide a per-flow/session traffic control ar the routers.

• Evaluation
  • Significance rating: 4
    QoS is an important topic in current development of the Internet, and it requests the explicit management of the resources. This framework presented in this paper supports both non-real-time and real-time communication in the extension of the current Internet. This paper concludes the current efforts on this topic, hence throws much light for the future research.

  • Convincing rating
    The authors describe the design principles of their architecture, then introduce the IS model and the implementation framework with some detailed considerations. The new architecture is compared with the old model, and the readers may get a clear big picture from this overview.
    One thing confused me is that how the situation will be whenever a working router suddenly crashed. Will it make a trouble for the IS network? Since it will lose the information, and the hosts it servicing will lose the path.

  • Limitation
    From the technical viewpoint, this model is still too complex to apply to the real networks, especially for the huge scale of the Internet.
    As for the methodology for this paper, the authors exclude other possibilities, e.g. designing a new infrastructure which is parallel to the current Internet. This narrows the focus down to find a good extension of the Internet, instead of looking for a fundamental revolution of the network infrastructure.

• Conclusion
  There is still a long way to go for improving the sevice types and quality of the Internet. However, one thing is clear that the research will dig through the internal of the network instead of focusing on the improvements at the end hosts and applications.

  11/5/01