Optimized Deployment Of Virtual Network Function In NFV

Network function virtualization(NFV)is a promising technique aimed at reducing capital expenditures(CAPEX)and operating expenditures(OPEX),and improving the flexibility and scalability of an entire network.In contrast to traditional dispatching,NFV can separate network functions from proprietary infrastructure and gather these functions into a resource pool that can efficiently modify and adjust network service.However,this emerging technique has some challenges.This thesis mainly studies two problems of network service deployment in NFV: reliable deployment of network services and response latency of network services.The reliability deployment problem proposed in this thesis is intended to ensure users' basic reliability requirements and their right to enjoy high-quality services while reducing operators' cost.Considering the high quality requirements of clients,this thesis firstly proposed a heuristic algorithm ER to guarantee the maximum reliability of the services.With a detailed study,this thesis found that the high reliability of the services will increase the cost and resource cost of operators.So,this thesis then put forward another heuristic algorithm ER_CS,which is based on ER,by reducing the reliability of the services(but not below the users' minimum requirements)to reduce the operators' CAPEX and OPEX.In this thesis,the performance of these algorithms are verified by using two arbitrary topologies of different sizes as the substrate network.The results of extensive experiments indicate that the proposed algorithms perform efficiently in terms of the blocking ratio,resource consumption,time consumption and the first block.The majority studies of network service deployment problem in NFV environment are about the deployment of service function chaining(SFC).And in order to reduce the response latency of network services,the majority of studies are limited to the traditional technologies,such as deploy virtual network functions(VNFs)onto the same substrate node thus saving transmission delay and bandwidth consumption.This thesis introduces a special mode of network service request,namely workflow service request,to solve the problem of high response latency of network services.This thesis then put forward an algorithm DMRT_SL which can dynamically minimum the response time of the workflow service request.DMRT_SL algorithm can help operators efficiently deploy the novel service requests onto the substrate network.At the same time,in order to take a contrast within the problem that whether the VNFs(which can be processed in parallel)can be deployed onto the same substrate node,DMRT_NSL is proposed to compare with DMRT_SL to illustrate it.The randomly generated topology is used as the substrate network;the two algorithms are implemented with the programming language of java and a large amount of simulation experimental data are obtained.The extensive results show that DMRT_SL algorithm is not only outstanding in the performance of average response latency,but also has some advantages over other excellent algorithm in terms of the blocking rate and time complexity.