Research On Mapping Mechanism Of Network Function Virtualization

With the rapid development of network technology,the demand of user for network services is more diversified.To meet the needs of user,netwrok operators need to continuously deploy proprietary network devices which contributes significantly expenditure for network operators.Network Function Virtualization(NFV)technology is a new method to overcome the defects of traditional telecom networks.NFV decouples network functions from proprietary hardware equipments,and flexibly deploy software-based Virtual Network Functions(VNFs)in network according to the demands of user.Traffics have to traverse through a sequence of VNFs in a pre-defined order,which is named as Service Function Chain(SFC).Since network operators always deploy the same type of VNF instances in different locations to make traffic flow can choose VNF instance and path flexibly.However,how to accommodate more SFC requests as far as possible to improve operators' profit becomes an important issue to be studied.Accordingly,this thesis researchs on the joint optimization of VNF selection and traffic steering in telecom networks to maximize the throughput of networks.For totally-ordered SFC requests steering problem,this thesis consider how to maximize traffic throughput while satisfying the end-to-end latency requirement and the resources constraint of nodes and links.This thesis first formulates the studied problem as an Integer Linear Programming(ILP)model,and then prove it is an NP-hard problem by reducing from the Virtual Network Embedding(VNE)problem.Due to the problem complexity,this thesis proposes an efficient heuristic algorithm based on dynamic programming to select VNF instance and steer traffic passing through the network,while satisfying the end-to-end latency requirement and the resources constraints of network.Simulation results show that the algorithm can effectively improve the throughput of networks.In order to adapt the algorithm to more scenarios,this thesis extends it to online version to handle SFC request dynamical arriving and leaving.For partially-orderedd SFC requests steering problem,this thesis considers how to maximize the throughput of network and satisfy the resource constraint when the VNF instance has an impact on the bandwidth of the traffic flow.To solve the problem,this thesis first formulates the problem as an ILP model.Since the problem is an NP-hard problem,an efficient heuristic algorithm named Partial-Ordered SFC Steering(POSFCS)is proposed to solve the problem.This algorithm separates VNF instance selection and link selection.It first constructs the partially-orderedd SFC requests as totally-ordered SFC requests according to the dependence of VNFs and the bandwidth impact of VNF instances.Then choose VNF instances by weighted bipartite graph matching.The node selection scheme is used as the input condition for link selection,and the randomization rounding method is used to obtain the link selection scheme.Simulation results show that the algorithm can effectively improve the throughput of networks.