Abstract
In recent decades, the demand for network services has grown rapidly. Applications and users make the network more complicated than it ever was. Services that require high bandwidth and/or low latency are greatly susceptible to the effects of traffic congestion. A novel network architecture referred to as Software Defined Networking (SDN) proposed in 2009 decouples the control plane from the data plane of switches, having one centralized controller for the network so that network managers could have chances to control and manage the network as desired. This paper presents a quality-of-service (QoS) management system integrated with a flow classification engine. This makes it possible for network managers to divide hosts and resources into several groups, for which the system identifies the optimal route with minimum costs, based on real-time monitored statistics and a pre-determined cost model. It also enables the alteration of routing paths in situations of traffic congestion. Finally, we adopted the concept of Collaborative Filtering for usage prediction and bandwidth allocation. Experiment results demonstrate the efficacy of the proposed algorithm and system architecture in providing a QoS with minimal overhead and precise bandwidth management. Finally, for achieving network virtualization and easy management, we also integrate the QoS system with a novel framework which realize the vCPE concept through SDN technology. Users or enterprise could just subscribe the QoS services and set the parameters which is needed and monitors the network status in real time via the web-based user interface we provide.