Cloud Computing-oriented Flexible Separation Theory And Method Of Control Plane And Data Plane

In the era of cloud computing, the large-scale deployment of virtualization and cloud computing spawn a next-generation data center network which centered on the workload. But running large-scale data centers will consume large amounts of energy. Researches show that the utilization of current data center is kept in 5% to 20% and a large number of servers are often in the idle state. However the power consumption of these servers is more than 50% of the full load condition. Thus how to run a data center efficiently is a very important research subject. How to use new ideas to solve the challenge of network under the new application? The Software defined Network (SDN) with control plane and data plane separation provides a new idea and new method for the people.Because of the resource scheduling and business requirements among the end entities, huge network traffic is emerged on network equipments and network links in the cloud computing network. The network traffic mainly consists of control traffic and data traffic. However, compared with the data traffic, control traffic only accounts for a small percentage. Due to the different requirements purpose, the paths in the cloud computing network went through by the control traffic and data traffic is not the same. But, in order to transport these traffic, related equipment and even all the equipment in the cloud computing network must be always running to ensure that the connectivity of the network. This is one of the main reasons for high energy consumptions, low efficiency of current networks, and it is also the main problem of current network architecture design. As a kind of network innovation architecture, SDN concentratedly analyze and control the whole network, and realize the function of network virtualization, and separate the control plane and data plane. These make the original router no longer precede route learning, but only be devoted to data forwarding. So it can control network traffic flexibility, and solve bandwidth bottlenecks of the cloud services when transmitting in the backbone network, in addition make cloud computing develop more widely.First of all, this thesis introduces the function of the control plane and data plane, network power consumption and enegy efficienct parameters of the control plane and data plane. Then study the theory of energy efficienct separation of control plane and data plane. As to control plane seperation which is based on energy efficiency priority, this thesis takes minimum control plane power consumption as the optimization objective. According to the abstract relations between control plane and data plane, this thesis sets up an integral linear programming (ILP) mathematic model of the control information of transmission network, and then puts forward the high failure recovery control plane energy efficienct separation algorithm (FRCS). In order to simplify the algorithm complexity, this thesis regards Hamilton optimal circle and algebraic connectivity of control plane network as the termination conditions for the algorithm to obtain the optimal control plane topology with the lowest energy consumption trough loop iterations.Secondly, as to the theory of energy efficiency priority data plane network seperation, this thesis aiming at the energy efficiency of data plane, adopting the modern graph theory to build the ILP mathematical model of the data information in transmission network, puts forward the max-flow transmission data plane energy efficienct separation algorithm (MMFDS). Contrapose the difficulty of multiple paths constraint of traditional algorithm, this thesis presents a path-orinted solution. And in order to improve the network energy efficiency of data plane, this thesis regards the path hop as constraint to establish flow transmission paths with minimum bits energy consumption for each end-to-end request to achieve high energy efficient data plane network seperation.Finally, as to the control plane and data plane network separation methods with the enegy efficiency priority, this thesis puts forward the control plane and data plane energy efficienct separation (EECDS) mechanism in cloud computing network through the collaborative energy efficienct transmittion of control plane and data plane network and the goal of minimizing the network energy consumption. According to the request of the network this algorithm sets up real-time control plane link utilization matrixes, switch off the low load link and redistributes the traffic which is transmitted by the low load link to ensure the transmission of data plane network traffic with trouble-free. Thus, this thesis set up the energy efficienct separation mechanism of control plane and data plane.