| Networks play an important role in today's highly connected world.The increase in the capacity of Internet access networks and backbone networks has brought changes in the traditional WAN market.The huge demand for cloud services means that there is a need for WAN connectivity,security and bandwidth.Great demand.The traditional network architecture is not suitable for meeting the cloud-based service requirements of modern data centers and operators.As an emerging network architecture example,Software defined network(SDN)is improving the scalability and flexibility of the network.Software-defined wide area network is a service formed by applying software-defined network technology to wide area network scenarios to construct,deploy and operate wide area networks.Segment routing(SR)defines the information path through the network through an ordered list of the multi-protocol label switching mechanism on the packet header on the ingress device.This system makes SDN routing management simpler and more effective.At the same time,because the amount of state information that needs to be configured in the core node to support complex services is greatly reduced,segment routing can also solve some scalability issues in SDN.However,in segment routing,multiple multi-protocol label-switched labels are stacked in the packet header to complete end-to-end transmission,which may result in a large label stack and a long packet header.Therefore,when segment routing is applied to a large network,scalability issues may occur.This article proposes two solutions to the above problems.First,this paper designs and implements a path optimization algorithm based on SR,and proposes a heuristic routing algorithm with bandwidth guarantee based on link/node centrality and link congestion index.Firstly,the link weights are calculated with the performance targets such as link centrality and link congestion index as a reference,and node weights are calculated with node centrality as a reference.Subsequently,the original network topology was updated with the calculated link weights and node weights.Furthermore,based on the updated network topology,this paper designs the traffic engineering based on centrality and link bandwidth(CLB-TE)path optimization algorithm.This algorithm can build a path with guaranteed bandwidth and minimize the demand for rejected traffic.Possibility.Finally,the proposed CLB-TE path optimization scheme is compared with other routing schemes,and the simulation verifies the performance advantages of the path optimization algorithm proposed in this paper.Second,due to the limitation of the maximum segment list depth,the limited label stack depth limits the ability of the source node to achieve a longer optimal explicit path.This paper solves this problem by improving the traditional segment routing coding algorithm.This paper first introduces the global Adj-SID,and combines the global Node-SID defined in the SR to encode and compress the original SR path,which effectively reduces the label consumption and the number of segment routing flow entries under the constraint of the maximum label stack depth.Next,for the long segment routing path,this paper designs an SR path coding based on additional paths(SRPE-AP)algorithm to further encode and compress the path,thereby minimizing the depth of the segment list.In addition,compared to traditional IP/MPLS networks that do not use label stacking,SR adds a segment list to each data packet,thereby introducing additional packet overhead caused by stacking multiple labels.In order to obtain the best path coding solution under the constraint scheme,this paper introduces data packet overhead as one of the optimization goals,and proposes the reverse build sub-path set(R-BSPS)algorithm to solve this problem.Simulations prove that the proposed mechanism and algorithmare superior to common segment routing solutions with the maximum segment list depth constraint,and the data packet overhead is also lower than existing solutions. |
PDF Full Text Request