Algorithms For Deploying Service Function Chain In Single Domain Networks

With the introduction of Network function virtualization(NFV),network service deployment can get rid of traditional dedicated hardware,improve the flexibility and agility of service deployment by sharing hardware,and save service costs.One of the key issues in the deployment of NFV service functions is the optimal deploying of service function chains(SFCs)in the physical infrastructure.Most of the existing researches are concentrated in a single domain,from how to allocate the underlying resources and ensure reliability,strict delay constraints in the deployment process,bandwidth limitations,etc.However,there is relatively little work on energy consumption.In actual production and life,the control of cost and expenses is very important.Therefore,in the deployment of SFC,it is of practical significance to study how to reduce the total energy consumption in the deploying process.In addition,there are few scholars involved in the scheduling of service functions when sharing physical hardware.The work of this thesis is still to address the unresolved issues in a single domain,and the thesis has studied the two issues mentioned above.First,the author studied the energy consumption problem in the SFC mapping process,and minimized the resource consumption in deployment while ensuring a certain successful deploymem ratio.The author analyzed the existing algorithms and found that the flaw of greedy method is the blindness of pathfinding,which may lead to a significant reduction in the successful deployment ratio.Therefore,the author put forward the idea of "direction guidance" and combined it with greedy method,and proposed the direction guided first fit(DGFF)algorithm and direction guided greedy(DGG)algorithm.The simulation results show that the two proposed algorithms are better than the existing ones.Secondly,the author also studied the service scheduling problem when sharing physical hardware.The process of sharing physical hardware can be divided into waiting queuing phase and comparison scheduling phase.In the waiting queuing phase,the author sorted the services that do not enter the shared part,and determined the order of service scheduling to ensure the fairness of the service and minimize the overall delay of all services.The author introduced the concept of dynamic priority and proposed a dynamic priority queue(DPQ)algorithm to solve the queue ordering problem in the queuing phase.In the comparison scheduling phase,the author can get the appropriate scheduling method by comparing the processing and transmission delay between different services to determine when to deploy SFC.Based on these mentioned above,the author proposed a comparing and scheduling method(CSM)to minimize the total processing delay in the comparison scheduling phase.The simulation results show that the algorithms we proposed are better than the existing ones...