QoS Architectures

The bottom-up mechanisms described so far can be considered building blocks in a larger architecture that describes how the Internet may deal with real-time traffic requirements. Two different architectures exist: Integrated Service and Differentiated Service. Both are based on common traffic policing strategies and can be combined in order to allow QoS optimizations both in local and wide area network environments.

Traffic Policing

Traffic policing describes the general concept of allowing traffic to enter the network only when certain conditions (such as sufficient resources to handle QoS requirements) can be met, as well as supervising traffic behavior. This is done to adapt the network to traffic requirements or to impose restrictions in cases of unfair resource usage.

There are different methods for traffic policing, ranging from fair queuing algorithms based on the QoS requirements within routers to the use of network-wide priority classes (which determine the handling of QoS-sensitive traffic within the routing/forwarding system) and adaptive mechanisms, such as random early detection by routers to probe traffic behavior and its sensitivity to artificially and randomly caused packet loss or delay.

Traffic policing (and its corresponding administrative and technical overhead) is used not only because of the QoS requirements imposed by real-time, multimedia traffic. There may be economic reasons, as well—for instance, to exercise cost control, such as when ...

Get IPv6 Essentials now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.