Analysis of QoS Schemes and Shaping Strategies for Large Scale IP Networks Based on Network Calculus

Abstract

IP network experts and engineers have been working on solutions for decades to promote the network QoS. Latency guarantee, as one of the key aspects of the QoS, is attracting increasing attentions with requirements from time-critical applications and the vision of building a fully connected, intelligent world. Meanwhile, Network Calculus is a theory that focuses on performance bound analysis for communication networks, and has been used in avionic networks. However, because of the extremely large scale and high complexity of IP networks, few works gave theoretically modeling and systematically analyzing for the QoS (i.e., latency bound) of IP networks. In this paper, three QoS schemes for IP networks are summarized and the performance on the perspective of efficiency is analyzed. The effect of ingress shaping is also investigated, and results show that a proper ingress shaping could benefit the overall network latency performance, and could be adapted to all three QoS schemes. An IP network use case is given with different QoS schemes applied and the performance is evaluated by using Network Calculus.