Research On Resource Allocation Mechanism Of Network Function Virtualization

With the rapid development of communication network,people are increasingly using the Internet in our daily life.These technologies bring convenience to the users at the same time also make the network many problems.Network Function Virtualization appears to be very in line with the current environment.Through NFV technology,the software implementation of the network function is decoupled from the dedicated hardware and migrated to the x86 server.NFV realizes the independent development of hardware and software,and provides dynamical and flexible network function deployment.In NFV network,SFC is a chaining set of ordered virtual network functions,which can be processed in sequence.However,the use of NFV is also facing some challenges.How to achieve efficient and flexible resource allocation is one of the difficulties that NFV deployment needs to solve.This thesis mainly studies the problem of NFV resource allocation.According to different user scenarios,the problem can be divided into two cases: known VNF sequence and unknown VNF sequence.In this thesis,corresponding NFV resource allocation schemes are proposed for these two cases.The problem of NFV resource allocation in the known VNF sequence is known as SFC deployment problem.In this thesis,the comprehensive consideration of computing resources and communication resources is used as the mathematical model to optimize the goal of the whole network resource.At the same time,this article makes the same function of different SFC as possible through the same VNF entity to reduce the waste of debris computing resources.The analysis shows that the single SFC deployment problem has hidden Markov property.Therefore,we design the Resource-efficient Virtual Network Function Placement algorithm(RVNFP)based on Hidden Markov chain and carry out simulation verification.The results show that the scheme can make computing resources and communication resources achieve a trade-off,which ensures that the overall resource consumption of the network is optimized while the network performance is also considered.In the unknown VNF order scenario,VNF has the attributes of bandwidth change factor and dependency relation.These characteristics make VNFs processing sequence and location have little influence on the total demand of network computing resources,and have a great impact on the total network link load.Based on this,our paper proposes a Traffic Aware and Interdependent VNF Placement algorithm(TAIVP)to minimize link load.The scheme consists of three steps: firstly,to build end-to-end SFC with the lowest cost of link;secondly,the path planning algorithm based on A* principle is used to search the feasible working path;finally,according to the influence of VNFs bandwidth change factor on link load,SFC is deployed to the working path to optimize the target function.The simulation results show that our scheme is better than the other algorithms in both the total link load and the request receiving rate.