Research On Rule Management In Software Defined Networks

With the arrival of mobile Internet era,all kinds of applications and services are growing explosively,and the demand for resources(including computing,storage,I/O,spectrum bandwidth,etc.)is also increasing.The traditional network architecture is no longer able to meet the growing demand.The proposed software-defined network(SDN)alleviates the shortcomings of traditional networks to some extent.SDN is a new type of network innovation architecture proposed by the clean slate research group of Stanford University.The core technology of SDN is that the control plane of the equipment is separated from the data plane,and a centralized controller is used to manage and configure various network equipment so as to achieve flexible control of the entire network communication and provide a good platform for the innovation of core networks and the development of applications.SDN breaks through the gap between the application program and the network,enabling programmers to autonomously control the network through programming interfaces according to application requirements.This is the future trend of network development.However,the software-defined network brings many problems which are worth studying when it meets the increasing application requirements and improves network performance.The existing research work on software-defined networks is mainly divided into two categories: one is to study the data forwarding problem at the data level,and the other is to study the interaction between the control plane and the data plane.On the one hand,existing data-level research in software-defined networks focuses on single-path routing.However,single-path routing protocols cause large amounts of overhead when there are medium or even large numbers of routes to be maintained;thus,multipath routing is ignored.It can improve the forwarding efficiency of the data plane and the advantages of the network resource utilization.On the other hand,when the network requirements change,the routing rules also need to be updated synchronously,but the existing work ignores the variability of the target network after the routing rules are updated.In this paper,a mathematic model is established for these two issues as a more complete work study.Because of the centralized control of SDN,the controller needs to periodically collect flow state information at the data plane.This paper aims to reduce the overhead of collecting flow state information and study the flow state information collection at the control level and data level under multi-path routing conditions.This article is divided into three studies.First,the first research content studies the optimization of routing rules in software-defined networks.By considering the flow balance and link capacity constraints,as well as the constraints of the TCAM(Ternary Content Addressable Memory),an average maximum throughput model that satisfies Integer Linear Programming(ILP)is constructed.In order to solve the computational complexity of the ILP problem,a heuristic algorithm with polynomial time complexity is proposed.Through a large number of experimental simulations,the effectiveness of the algorithm is proved.Second,based on the first research content,the second research content discusses the update of routing rules at the data level.With the help of the virtualized network environment,the impact of different virtual network function(VNF)migration strategies on the update process of routing rules is analyzed.The update of routing rules is abstracted into two sub-problems: the VNF migration problem and the update of routing rules.By jointly optimizing migration overhead and update delay,an optimal VNF migration strategy and routing rule update model is constructed,and this problem is modeled as a Mixed Integer Non-Linear Programming(MINLP)problem.A corresponding heuristic algorithm is proposed,and the effectiveness of the algorithm is proved by a large number of experiments.Finally,the third study is based on the interaction between the control plane and the data plane.When collecting flow state information,selecting different switch sets to collect flow state information will generate different costs.The routing policy of control information will directly affect the selection of the switch set.In addition,in the case of multi-path routing,certain redundant information is unavoidably generated in the collection process.Selecting the optimal switch set will reduce part of the overhead.Therefore,we study the minimization of the total overhead generated during the collection of flow state information as a goal,and reasonably select the routing-based switch set.A set coverage model that satisfies ILP is constructed.Finally,a two-phase heuristic algorithm is proposed.Compared with the traditional single-path routing algorithm,the experimental results show that the algorithm is effective in energy conservation.The contribution of this article is:A comprehensive review of current work and research progress in software-defined networks,including the evolution of architecture from traditional networks to software-defined networks,the application of software-defined networks in traffic engineering,and the enhancement of network performance in software-defined networks Research etc.The first study took into account the application of multipath routing in the routing rule placement optimization problem in software-defined networks,and considered the flow balance,link capacity limitations,and TCAM capacity requirements to model the problem as a ILP problem with multiple constraints.The second study took into account the impact of VNF migration on the update process of routing rules in the context of virtualized network functions.This problem was effectively solved by minimizing the migration costs of VNFs and the update delay of routing rules.In the third part of the study,the interaction overhead between the control plane and the data plane of the software-defined network is considered.The flow state information collection model under the condition of multi-path routing is studied,and the collection strategy with the lowest comprehensive overhead is proposed.At last,we verify the feasibility and effectiveness of the program.