A Service-Oriented Load Balancing Mechanism For Software-Defined Networking

With the rise and development of the services of Mobile Internet, cloud computing and big data, the scale of network and traffic are increasing exponentially. At the same time, due to the enriching of the new Internet applications constantly, the quantity of services(QoS) required by subscribers is changing dramatically. Therefor, how to balance the network load and guarantee a certain QoS are now highly priorized problems to be solved. But the traditional Internet has become increasingly difficult to meet the current development needs. Software-Defined Networking(SDN) is an emerging network paradigm which decouples the control plane from the data plane, reduces the complexity of network management and has global network status view. It can simplify the network flow control and guarantee QoS flexibility, but also faces with the scalability issue. Therefore A Service-Oriented Load Balancing Mechanism for Software-Defined Networking is proposed to resolve the problems of network load imbalance and the scalability in control plane. The main research works includes:Firstly, a load balancing model is proposed for distributed controllers. In the model, the the flow-requests information among the switches is used as the basic unit of the controller, all controllers publish respective the total of the flow-requests periodically and flow-requests deviation mean is introduced to aware load status for controllers. On this basis, a load balancing algorithm is proposed based on load-aware. This algorithm uses flow-requests information allocation strategy, taking into account the current load and propagation delay of idle controllers and allocating partial the flow-request information of the overload controller to the idle controller with the minimum of flow-requests deviation and propagation delay. In order to avoid inconsistency of the network status caused by performing load balancing algorithm at the same, each controller maintains a flow-request deviation the mean table and the overload controller performs the load balancing algorithm based on the magnitude of the value in the table. Experimental results show that the model can adjust the load of controllers quickly and effectively, has lower time complexity and better robustness.Secondly, service-aware adaptive link load balancing mechanism is designed which can aware the service types via the entended the northbound interface and monitor the state of the network periodically. Adaptive link load balancing algorithm is proposed and the link weight is introduced based on QoS-aware, which measures the comprehensive quality of link throuth acquiring real-time QoS parameters. So the mechanism can select the path of the best link quality in the current to forward data, reduce the imbalance of the load distribution in the network. Meanwhile for different service types, dynamic QoS routing optimization strategy is proposed further. It adopts the Lagrange relaxation technique to compute a path which meets the QoS constraint. Experimental results show that the proposed mechanism can effectively balance the network traffic, the average utilization of link bandwidth up to 79%, and achieve QoS guarantee for different services.