How DCN and SDN meet new requirements New Technology of ip ran Network

How DCN and SDN meet new requirements New Technology of ip ran Network

The traditional 3G/B3G network needs the basic bearer network to provide the Layer 2 capability to meet the needs of the Upper-layer network mobile service return. However, due to the introduction of S1-Flex and X2 interface, the basic bearer network must have three-layer switching capabilities. Based on the dynamic IP/MPLS protocol and key technologies, the ip ran is a solution that combines Layer 2 and Layer 3 technologies to meet the requirements of base station backhaul. The following describes the key requirements and related technologies of the ip ran network.

High Availability networking principles

In order to control the service convergence time when a network failure occurs in milliseconds, the network must support some advanced fast convergence technologies and organically integrate these technologies. Typical fast convergence technologies are as follows: IGP fast convergence technology, IGP fast re-routing technology, priority routing convergence technology, BGP next hop tracking trigger (NHT) BGP next hop Separation Technology and prefix-independent convergence technology (PIC.

Priority routing convergence technology is an important technology to improve network reliability, which can provide faster route convergence for key businesses. To ensure rapid changes in BGP Loopback and accelerate convergence of VPN routes in an mpls vpn environment, priority routing convergence technology must be enabled to increase the priority of IGP's 32-bit Host Routing, accelerate bgp nht tracking.

The BGP prefix-independent convergence technology (PIC) achieves on-demand iterative update through multi-layer separation of the bgp fib table, ensuring that the multi-path, the fast convergence of VPN routes is decoupled from the number of VPN routes.

Fast Convergence of vpn frr and vpn ecmp is mainly used to detect faults in the remote PE path. There are two methods: BFD for Loopback (Peer BFD) method and NHT method (by tracking the 32-bit loop back address of the remote PE in the igp rib table ). NHT is also a good choice because the BFD method has problems such as interconnection, tedious configuration, and quantity and specifications. However, to ensure a certain convergence speed, NHT needs to combine the PIC (Next Hop separation) and IGP fast convergence technology. When the 32-bit loopback address of igp rib is found to be faulty, update the content of the next hop pointer of the VPN or set the RR table of the vpn frr. For vpn ecmp, you can use the method of deleting the next hop or setting the next hop to an invalid one to perform fast rerunning.

If you locate the actions to update the route table at the intermediate layer, the BFD method can be considered as a bottom-up NHT technology, while the bgp nht is a top-down NHT technology. The two work flows are basically the same, the main difference is that the BGP NextHop detection method and detection speed. The two have little impact on the number of VPNs and the number of routes. Even if the two are affected, there is no big difference in the degree of impact between the two. The key technology here is to adjust the NextHop table of BGP Multi-path immediately when NHT occurs, and there should be a corresponding delay iteration time (5 s by default) for VPN routes ), that is, the bgp fib table (FRR or Fast Ecmp) is updated on demand first, and the entire bgp rib table is converged after the Protocol process.

Efficient O & M principles

The features of the LTE band determine that the deployment of eNodeB for LTE base stations tends to be intensive. To achieve better coverage, eNodeB must multiply the number of 3G base stations. For example, the number of 3G base stations (including macro stations, micro stations, and office stations) in a developed domestic network reaches 2000 ~ With 4000 servers, the eNodeB station has approximately 8000 servers. Such a large number of 3G and LTE base stations will cause a series of thorny problems.

More than 10000 IP/MPLS base station bearer devices will be added to the deployment of massive base stations. The introduction of massive IP/MPLS devices will bring unprecedented complexity to network management, and the distribution of devices poses unprecedented operational pressure on operators. Second, there are multiple protocols between the network access device and the aggregation device, which require high technical capabilities of network O & M personnel and complicated network operation and maintenance, putting pressure on the operator's OPEX.

The current ip ran network is based on the traditional general-purpose router platform. It constructs an ip ran network with a large number of devices through complex IP/MPLS routing protocols, which brings great challenges to the deployment of high-availability networks. Especially when the network is fluctuating, it is a challenge for devices in the network, as well as for network reliability and fault location.

The aggregation nodes need to be configured more frequently for service activation, and the IP/MPLS technology configuration is complex. The command line configuration method of traditional router equipment is not applicable to the configuration and adjustment of large-scale network nodes, in addition, there are many configuration parameters, complicated service configurations, and difficult to locate in case of service faults, resulting in a sharp increase in Service Activation complexity and maintenance costs.

With the development of technologies and business applications, new value-added services and applications are constantly emerging. How can the ip ran network better adapt to the development and requirements of the business network, rapid response to service establishment and Dynamic Maintenance of services are also one of the problems currently encountered by the ip ran network.

Ip ran solution providers and operators jointly research and discuss these issues, and propose the next-generation ip ran network technology. At present, there are two main technologies in the evolution process: DCN self-Connection Technology and SDN-Based ip ran network virtualization technology.

DCN self-Connection Technology

For the deployment of a large number of base stations, on-site debugging and on-site configuration services are time-consuming, resulting in a huge workload. When business adjustment is frequent or network expansion is urgently needed to cope with business expansion, on-site business deployment efficiency is low.

The DCN self-pass solution solves the difficulties of on-site configuration and on-site commissioning. After using the DCN self-connection solution, the access layer device can automatically connect to the management channel of the network management system after power-on, and then perform a series of maintenance and business configuration operations on the device from the network management system.

The DCN self-Pass scheme requires that the IP address of the interface between the A-A and the A-B device is not configured, using the unnumbered Ethernet interface (that is, the interface does not have an IP address, the IP address of only other interfaces can be used for communication ). When the A-A/A-B device interface communicates for the first time after power-on, it uses the loopback loop address automatically generated by the device itself, and spreads the route information LSA Based on the point-to-point network that runs OSPF protocol, devices A and B automatically learn these diffusion routes to form A route table, so that the DCN channel can communicate with each other.

The IP address of the DCN loose Lo-xx can be automatically allocated or manually specified (for backend address re-planning.

The DCN self-connection solution requires that A/B device solidify some information when leaving the factory, including: lo-xx interface, Lo-xx address generation rule, DCN Vrf, DCN-specific OSPF process, and VLAN of the LSA sub-interface used to forward the OSPF diffusion.

After the power-on of the device, the DCN self-pass Solution passes through a series of automated steps. The routes of all devices in the entire access ring are updated, and the updated A-B topology is formed on the network management system, manual on-site configuration is not required, which greatly improves the efficiency of device installation and commissioning.

SDN-based network virtualization technology

After SDN is introduced, ip ran network control plane and forwarding plane are separated. Access Device A and aggregation Device B are integrated with the control proxy module for protocol interaction between the forwarding plane and the control plane; device B retains the original routing protocol and other functions for L2/L3 VPN communication with devices in the man.

Forwarding plane: considering that the current network has been deployed on a large scale, in order to be compatible with the existing network hardware conditions, at the initial stage, the network meta-Level Interconnection and data stream forwarding must be compatible with the IP/MPLS protocol set (the MPLS-TP is its subset ). Later, the open flow table chip will gradually evolve to the Open Flow flow table forwarding.

Control Plane: defines the control plane at the business level. Based on the openflow protocol, the control plane is extended to discover the network topology, distribute service configurations, calculate the business PW/LSP path, and distribute table items. Consider supporting multiple protocols in the future, such as I2RS.

Management plane: the forwarding device still has an independent management plane, which can be used as an independent network element for management on the network management system. However, because the business has been centrally controlled by the Controller, the Network Element Management area only provides the Network Element device management function. The Controller provides network-based northbound interfaces for network management and third-party applications.

SDN-Based ip ran network virtualization technology enables efficient O & M management and high software automation. Some existing problems, such as clock synchronization and L2/L3 multicast, will also find a good solution during the rapid development and application of technology. SDN-Based ip ran network virtualization technology and flexible and rich application implementation capabilities are considered as the future development direction of ip ran network.