The purpose of this paper (memo) is to discuss the proposed extension of the Internet architecture and protocols to provide integrated services (support for real-time and non-real-time services of IP).
The main contribution of this memo is to present an overview of the elements necessary to extend the Internet architecture to support integrated services.
(1) Real-time applications often do not work well across the Internet
because of variable queueing delays and congestion losses.
(2) The real-time QoS infrastructure required for real-time applications must be designed from the beginning for multicastinging; simply generalizing from the unicast (point-to-point) case does not work.
(3) The proposed architectural extension is comprised of two elements: (1) an extended service model (IS model), which defines the externally visible behavior, and (2) a reference implementation framework.
I would rate this paper as a 4 because it clearly outlines the design space in which to build an extension to the existing Internet architecture which supports integrated services.
As opposed to most papers which we've read for this course, this paper presents an outline of the design landscape for this problem as opposed to performing specific experiments, simulations, etc. The authors did implement a code package (CSZ) which realizes the services described in the memo.
One limitation of the authors' approach is that I felt like the authors were forced to make some rigid assumptions in order to make this document limited in size. I wonder how relaxing some of the assumptions used by the authors would change the design space for this problem.
One lesson researchers should take away from this work is that one doesn't need to start entirely from scratch when building additional functionality into a system.