Optimizing QoS Partition based on Threshold-crossing Theory

With the advent of various QoS technologies, various traffic classes with different QoS requirements are supported on an integrated IP network. As for the multiservice IP networks, the dominant factor in the cost of provisioning a particular service is dependent on the minimum network bandwidth allocated in the network with QoS requirements satisfied. Therefore, to optimally partition bandwidth resource, it is necessary to deduce the QoS guaranteed minimum network bandwidth for each service. Currently, the administrators always prepare QoS partition plan via experience. This usually leads to large amount of waste resources and unnecessary budget. Since network traffic data is proved to be self-similar, it is suitable to analyze the previous network data of a particular service to deduce the minimum network bandwidth of the service. In this paper, Threshold-crossing Theory based QoS Partition Optimization Method is put forward to determine the minimum network bandwidth of a particular service based on its previous network traffic data in production environment. This method can be applied into optimizing QoS partition when designing, upgrading and partitioning network bandwidth and provide basis for reducing resource waste and avoiding unnecessary investment by optimizing all available network bandwidth.