Modelling and performance analysis of application-aware resource management

Application-aware resource management is the approach to tailor access networks to have characteristics beneficial for the running applications and services. This is achieved through the monitoring and integration of key performance indicators from the application layer within the network resource management. The aim is to increase user-perceived quality and network resource efficiency by traffic engineering with the help of these indicators. Using analytic and simulative approaches, this paper provides analysis methods for network operators to quantify the performance gains of alternative resource allocation algorithms that implement the application-aware concept. Network operators can use the proposed methods to evaluate possible performance gain trade-offs between investing in a pure capacity increase (over-provisioning) and the realization of an application-aware resource allocation. For this purpose, we model and analyse the application quality trade-offs of four algorithms for application-aware resource management at a single link in varying traffic situations. The algorithms are chosen with respect to different complexity and implementation level in order to cover the design space in a systematic way. The study of the algorithms focuses on the application-layer performance for the most used applications today, namely, web browsing and video streaming with constant bit rate as well as hypertext transfer protocol progressive streaming with variable bit rate. Application quality trade-offs are analysed in particular for a high-resource utilization at a bottleneck link. The results confirm that application-aware resource management outperforms best-effort resource management in terms of quality of experience. Moreover, our study provides guidelines for the selection and configuration of the evaluated algorithms.