Paper review: Integrated Services in the Internet Architecture: an Overview

Reviewer: Kevin Hofstra

  1. What is the best way to provide real time services over the internet?  Is best effort transport reliable enough to provide adequate services for time sensitive packets?  How can we differentiate between packets that must have a higher QoS, and what should the method of payment be?
  2. A standards proposal for a method of providing quality of service over the current internet.  An evaluation of how this improves upon previous methods and why it was designed in this particular way.
  3. A.  Traffic Control mechanisms actively supported:

i.                     Guaranteed Delay bounds.  Because of the WFQ scheduling there always exists an upper bound for delay that can be assured.  This is important for interactive applications. 

ii.                   Link Sharing Allows multiple distinct shares to be active on one link and then allocates additional space as free bandwidth.

iii.                  Predictive real-time service.  Allows separation of traffic that has different service objectives and leads to the optimization of both.

  1. Paper consists of 2 elements:  An extended service model (IS model) that provides the details of the added features of Integrated services, and a reference implementation framework that defines the types and properties of the different pieces of the IS model.
  1. Critique the main contribution
  2. System researchers and builders should recognize that quality of service over the internet is a must, even back in 94.  As more traffic needs to be real-time there must be a way of deciding who gets priority and how it must be done.  This paper is rather broad, and although its ideas have not been tested, it is bold in its attempts to create a service model without much additional overhead.  You must realize that at this time, creating a 2nd internet for real-time services was seen as an option, and with proposals such as this, it was know that would not be necessary.