Pezaros, D. (2005) Network traffic measurement for the next generation internet. PhD thesis, Lancaster University.
Abstract
Measurement-based performance evaluation of network traffic is a fundamental prerequisite for the provisioning of managed and controlled services in short timescales, as well as for enabling the accountability of network resources. The steady introduction and deployment of the Internet Protocol Next Generation (IPNG-IPv6) promises a network address space that can accommodate any device capable of generating a digital heart-beat. Under such a ubiquitous communication environment, Internet traffic measurement becomes of particular importance, especially for the assured provisioning of differentiated levels of service quality to the different application flows. The non-identical response of flows to the different types of network-imposed performance degradation and the foreseeable expansion of networked devices raise the need for ubiquitous measurement mechanisms that can be equally applicable to different applications and transports. This thesis introduces a new measurement technique that exploits native features of IPv6 to become an integral part of the Internet's operation, and to provide intrinsic support for performance measurements at the universally-present network layer. IPv6 Extension Headers have been used to carry both the triggers that invoke the measurement activity and the instantaneous measurement indicators in-line with the payload data itself, providing a high level of confidence that the behaviour of the real user traffic flows is observed. The in-line measurements mechanism has been critically compared and contrasted to existing measurement techniques, and its design and a software-based prototype implementation have been documented. The developed system has been used to provisionally evaluate numerous performance properties of a diverse set of application flows, over different-capacity IPv6 experimental configurations. Through experimentation and theoretical argumentation, it has been shown that IPv6-based, in-line measurements can form the basis for accurate and low-overhead performance assessment of network traffic flows in short time-scales, by being dynamically deployed where and when required in a multi-service Internet environment.