Energy-Saving Routing Strategy In Software Defined Data Center Networks

With the development of cloud computing and the increasing demand for video data business,the number and size of data centers increase dramatically,so that the energy consumption of data centers increases.However,the huge energy consumption of the data center limits the development of the data center,so the energy saving of data center becomes a hot topic.Numerous work has studied the issue of network energy conservation in data centers.On the one hand,from the dimension of time,some literatures put forward the exclusive routing algorithm to avoid link sharing and shorten the transmission time of the flow.On the other hand,from the power dimension,some research efforts can save energy by finding a set of network devices that are as small as possible to satisfy the traffic request and setting unnecessary network devices to sleep.The proposed algorithms reduce the network energy consumption to some extent.However,they only studied network energy consumption from a time or power perspective,ignoring device activation energy consumption and rule installation energy consumption.To solve these problems,the research contents of this thesis are as follows:(1)This thesis jointly optimizes power and time to minimize network energy consumption.We first analyzes the characteristics of switch and network energy consumption in the network,constructs the minimum network energy consumption(MNEC)problem,and analyzes the complexity of the problem,which proves that the MNEC problem is an NP-hard problem.The exact solution to the NP-hard problem is time-consuming,and the flow in the data center is very sensitive to delay.Therefore,we proposes an ITP algorithm with low time complexity,which combines the link sharing avoidance algorithm and the switch aggregation algorithm.The switch aggregation algorithm aggregates traffic on as few switches as possible,reducing the number of switches used and reducing power.Thelink sharing avoidance algorithm allows the flow to occupy the link bandwidth resources separately by avoiding the sharing of the bottleneck link,thereby shortening the time of flow transmission.Thus,the ITP algorithm combines power and time to reduce network energy consumption.In the experiment,we compare the energy-saving performance and solution time of the ITP algorithm and the optimal solution greedy algorithm.The results show that the energy-saving performance of the ITP algorithm can be close to the optimal solution.Compared with the optimal solution greedy algorithm,the ITP algorithm can greatly shorten the solution time.Compared with the latest two network energy-saving algorithms,the ITP algorithm can greatly reduce network energy consumption and shorten flow completion time.(2)In order to further reduce network energy consumption,this thesis considers network energy conservation from the perspective of flow,and divides network energy consumption into data transmission energy consumption,rule installation energy consumption and device activation energy consumption.Previous studies have only optimized the data transmission energy consumption,ignoring the device activation energy consumption and the rule installation energy consumption.However,device activation energy consumption and regular installation energy consumption account for a significant portion of total energy consumption and cannot be ignored.Firstly,the characteristics of flow energy consumption are analyzed,and a joint optimization problem of device activation energy consumption,rule installation energy consumption and data transmission energy consumption is constructed.It is proved that the problem is NP-complete.Subsequently,we propose a GN algorithm,which generates a weighted graph for each flow and finds the minimum network energy consumption path by finding the minimum weighted path in the graph.In the experiment,we compare the energy saving gap and time gap between GN algorithm and the optimal solution solved by Gurobi.The experimental results show that compared with the optimal solution,GN algorithm can approach the optimal energy consumption performance very well,and can greatly shorten the calculation time of path planning.We compare GN algorithm with two latest energy-saving algorithms.The results show that GN algorithm can greatly shorten the completion time of each flow and reduce network energy consumption compared with previous algorithms.