Core-stateless traffic control mechanisms for high speed networks.

摘要

In recent years, the trend toward wider use of digital multimedia has characterized the field of computer networks. The explosive growth of web traffic in the Internet is now being matched by the explosive growth of streaming media traffic. Existing streaming media applications not only generate large volumes of traffic, but also they are not as responsive to network congestion as TCP-based web traffic. As a result, existing streaming media traffic can cause two major maladies in the Internet: congestion collapse, and unfair allocations of bandwidth among competing traffic flows.; A promising approach to mitigating these maladies is to enforce all types of applications, multimedia and data, to employ a source-adaptive transmission mechanism that responds to congestion feedback from the network. This dissertation provides a comprehensive framework for multimedia and data applications to coexist in a friendly manner and to perform their best. A set of scalable core-stateless mechanisms comprising source-adaptive algorithms, network enforcement and fair queueing mechanisms are introduced in order to optimize the performance of multimedia and data applications, prevent congestion collapse and promote fairness in the Internet. In particular, this work introduces the concept of Source-Adaptive Multi-layered Multicast algorithms and the Network Border Patrol congestion avoidance mechanism. The main impact of the core-stateless traffic control mechanisms proposed in this work is in enabling the Internet to operate free of congestion and in a fair manner even in the presence of unresponsive multimedia flows. Furthermore, by providing applications and network policers with up-to-date congestion feedback information, the proposed mechanisms create an Internet environment where multimedia and data traffic coexist in a friendly and scalable fashion.