Research On Bandwidth Allocation Algorithm In Elastic Optical Data Center Network

In recent years,cloud computing technology has brought great changes to the operation form of enterprises,and more and more users are transferring their computing needs to cloud.However,with the increasing number of cloud users,the support of Data Center(DC)with the advantages of high-performance computing and large-capacity storage is even more needed.Besides,a large number of virtual machines are deployed inside the DC to ensure frequent communication between users.At this point,tasks submitted by users are assigned to multiple servers for collaborative execution to improve the completion efficiency of tasks,which requires the construction of an effective data center network to ensure efficient communication between servers.Among them,the traditional Data Center Network(DCN)switching mode is realized through electronic switching.Then,with the continuous increase of Data traffic,this switching mode can no longer effectively meet the demand of high bandwidth for emerging applications.For this reason,DCNs will adopt more optical interconnection technology to provide high bandwidth capacity.By the way,the traditional Wavelength Division Multiplexing(WDM)optical transmission network has received more attention in the early stage.But in fact,due to its inability to allocate more fine-grained spectral resources for cloud user connection requests,the bandwidth utilization rate is reduced.Fortunately,the Optical Orthogonal Frequency Division Multiplexing(O-OFDM)technology can divide the bandwidth into multiple orthogonal subcarriers so as to accomplish the flexible sharing of bandwidth in a more fine-grained way.Therefore,the Elastic Optical Networks supported by O-OFDM technology has been widely used in DCs.Besides,Elastic Optical Data Center Network(EODCN)provides users with large subcarriers,but there may be multiple users applying for the use of these subcarriers at the same time,so how to effectively allocate the finite subcarriers to user requests becomes particularly important.Based on this,this paper will focus on the bandwidth allocation in EODCN.In this paper,the bandwidth allocation problem studied in EODCN is mainly to achieve two goals.Firstly,it will comprehensively consider the benefits of cloud service providers and the performance of user connection requests,with the goal of maximizing the benefits of cloud service providers,to study the issue of bandwidth allocation.Secondly,the problem of user request delay is studied when the bandwidth resource is limited and the demand for bandwidth is increasing continuously.Finally,effective bandwidth allocation algorithms are proposed to improve the utilization of bandwidth resources.The main research work of this paper is as follows:1?Aiming at the bandwidth allocation problem in EODCN with the goal of maximizing revenue,since the subcarriers must meet the continuous allocation requirements,the first consideration will be to set the interval frequency band during the subcarrier allocation to avoid signal interference.After the carrier,the concept of virtual subcarrier is added,and then a profit model is established to calculate the profit.Secondly,this paper will combine the revenue model and propose an optimal bandwidth allocation algorithm based on dynamic programming.The algorithm can obtain the optimal subcarrier allocation plan through recursion to achieve the maximum benefit.Finally,from the perspective of improving the time efficiency of the algorithm,a heuristic-based near-optimal bandwidth allocation algorithm is designed,which can effectively reduce the algorithm time complexity and obtain near-optimal benefits.2?Aiming at the bandwidth allocation problem in EODCN with the goal of minimizing delay,the delay of user requests is often caused when subcarriers are allocated for the increasing number of user requests under the limited bandwidth.For this reason,two goals regarding delay guarantee are put forward,one is to minimize the sum of user request delays,and the other is to minimize the maximum user request delay.Then,this paper will design delay calculation models from two goals and model them in turn.Besides,the two models will use the concept of interval frequency band and virtual subcarrier to avoid signal interference.Firstly,aiming at goal one,on the basis of the delay calculation model designed for it and consider designing an algorithm based on dynamic programming.However,due to the high time complexity of the algorithm,this problem is transformed into a minimum cost network flow problem and a capacity scaling algorithm is proposed,which can achieve the optimal solution in polynomial time.Secondly,a bandwidth allocation algorithm based on dynamic programming to minimize the maximum user request delay is proposed for the second goal.Finally,in order to verify the effectiveness of the algorithm proposed in this paper,a heuristic-based minimum delay weight bandwidth allocation algorithm is proposed as a comparison to participate in the experiment.3?In this paper,the proposed algorithms are simulated and the results are analyzed.Aiming at the bandwidth allocation algorithm designed to maximize revenue in EODCN,we compare the two proposed algorithms with the improved existing algorithm to verify the effectiveness of the proposed algorithm in maximizing the revenue of cloud service providers.Then,aiming at the bandwidth allocation algorithm designed to minimize the delay in EODCN,we compare the performance of the algorithm designed under different goals with the benchmark algorithm to verify the effectiveness of the proposed algorithm.