Implementation of OSPFv3 dynamic routing protocol on NBMA Link

1. Question proposal
With the advent of the network information age, Internet technology has developed rapidly worldwide. Open Shortest Pass First (OSPF) dynamic Routing has become the preferred internal gateway protocol for network routing.
　　
In the transition from IPv4 to IPv6, OSPFv3 (Open Shortest Path First version 3rd), as the Core routing technology in the Next Generation Network, has attracted enough attention and in-depth research from academia, however, the current research and implementation mostly run on the network topology composed of routers connected to serial lines, while the Internet includes multiple subnet technologies (that is, data link technology ): such as Ethernet, 802.5 ring network, FDDI ring, Frame Relay subnet, ATM, and group wireless network. Therefore, it is necessary to study the implementation mechanism of the OSPFv3 protocol on multiple links. This article will study the implementation scheme of the OSPFv3 protocol based on the NBMA link type.
　　
2. Basic principles of OSPFv3
The Open Shortest Path priority (OSPF) algorithm is a routing protocol specially designed for IP networks by the Internal Gateway Protocol workgroup of the Internet Engineering Task Group (IETF, it is a region-based dynamic routing protocol based on the Link State algorithm and Dijkstra algorithm. OSPFv3 is version 3rd of the Protocol and is a mainstream protocol for routing technology in IPv6 networks.
　　
2.1 discovery and persistence of neighbors
　　
OSPFv3 uses the call protocol to discover neighbors and maintain the adjacent State. The call protocol is implemented by regularly sending Hello protocol packets. Each interface running OSPF sends a Hello packet. The Hello packet is used to discover and maintain the neighbor relationship, and finally forms an Adjacency ).
　　
On the Broadcast (Broadcast) and non-Broadcast Multi-Point Access NBMA (Not Broadcast Multicast Access) networks, Hello is also used to select the designated Router DR (Designed Router) and Backup assignment Router BDR (Backup Designed Router ).
　　
2.2 Database Synchronization
　　
Each vro running the OSPFv3 protocol maintains a Link status DataBase (LSDB: Link State DataBase) used to track the network Link status ). The database stores various Link status notices (LSA: Link State Advertisement) that reflect the status of routers and their links. These statuses include available router interfaces, known accessible routes, and status information of each Link. The link status database is actually a complete network ing graph for the region. It is the basis for the router to establish a route table. A vro in an autonomous system must form a consistent view of the network topology, they must be synchronized to their link status database. OSPFv3 relies on the "adjacent protocol", "Exchange Protocol", and "diffusion protocol" to complete data packet interaction, and ultimately implements LSDB synchronization for all routers in the same routing domain.
　　
2.3 Working Principle of OSPFv3 on NBMA
　　
OSPFv3 classifies different subnet technologies into the following categories: point-to-point subnet, Broadcast subnet, NBMA subnet (Non-Broadcast Multi-Access, Non-Broadcast Multi-Channel Access subnet), and point-to-Multi-Point subnet. The point-to-point link refers to the link connecting a pair of routers, such as a 56 KB serial line. The broadcast link has the broadcast capability, and each pair of routers on it can communicate with each other directly, such as Ethernet; the NBMA link does not have the broadcast capability and requires that routers on the two routers can communicate with each other directly to simulate operations on the broadcast link, such as the ATM network; A point-to-multi-point link refers to a set of non-broadcast links, such as frame relay.
　　
OSPFv3 has the following features on the NBMA link:
　　
(1) Neighbor Discovery Mechanism: Because the NBMA network type cannot automatically discover a neighbor, you need to manually configure the neighbor and use the Neighbor Discovery Mechanism unique to the NBMA to query the neighbor.
　　
(2) packet sending method: the broadcast link and point-to-point link state update package are sent to the multicast address, while NBMA does not have multicast capability, so the link state update MUST be copied, then, send the message to each neighbor router.
　　
(3) Pan-flood Mechanism: The NBMA link is the same as the broadcast link. After the DR (specified router) and BDR (Backup specified router) are elected, all other routers interact with DR and BDR databases.
　　
3. Implementation of OSPFv3 on NBMA Link
Because the NBMA network type cannot automatically discover neighbors, you need to manually configure the neighbors, including the local address of the neighbor's IPv6 link and the neighbor's Router-ID. First, set an interface of the router to the NBMA link type, and configure the neighbor on this interface.
　　
According to the working principle analysis implementation process: After the neighbor is manually configured, the neighbor is added to the neighbor list of the interface, according to the normal rules for sending the hello packet, at each interval, HelloInterval will send a hello packet to the neighbor. In RouterDeadInterval, the neighbor must receive a response from the neighbor to think that the neighbor can communicate with each other to start DR and BDR election, then, perform the following operations. If the RouterDeadInterval does not receive the hello packet response from the neighbor, the neighbor is considered a false neighbor and the neighbor is deleted; (generally, when a neighbor is automatically discovered for a broadcast link, such as a RouterDeadInterval, the neighbor is automatically deleted from the interface list if no hello response is received; in the NBMA link type, if there is no neighbor response in RouterDeadInterval, The Inactivity_timer event is triggered to switch the neighbor to the Down state, and the query neighbor timer (send the Hello packet) is enabled ), the timer interval is PollInte. Rval. If the neighbor response is received (if the neighbor is activated, the reply_timer timer is enabled to send reply), it indicates that the neighbor configured manually exists, and the poll_timer query neighbor timer is disabled; if no response from the neighbor is received within the query time, the poll query continues until the neighbor is manually deleted ).
　　
3.1 Implementation of the Neighbor Discovery Mechanism
　　
The neighbor of NBMA finds that it depends on the poll_timer timer to query the existence of the neighbor until the neighbor starts the reply_timer timer to respond to the neighbor, so that the neighbor is found. At the same time, the poll_timer timer is disabled.
　　
(1) query neighbor Module
　　
This module is used to manually configure neighbors and query neighbors based on the NBMA Neighbor Discovery Mechanism.
　　
(2) receive Hello Packet Processing Module
　　
This module is used to process the received Hello packet and perform NBMA-related operations based on the judgment of the neighbor information in the received packet.
　　
(3) Send Hello Packet Processing Module
　　
This module is processed after finding the neighbor, so it does not involve a series of NBMA-specific operation mechanisms such as poll_timer and reply_timer. You only need to determine whether the interface is of the NBMA type when sending the Hello packet, and send the Hello packet according to the judgment.
　　
3.2 Implementation of Database Synchronization
　　
The Database Synchronization Method of the NBMA network is the same as that of the broadcast network. First, select the specified vrodr DR and backup the specified vro BDR. All other vrouters begin to perform database exchange with the DR and BDR, in addition, flood traffic to the NBMA network always reaches other routers connected to the NBMA network through DR.
　　
The only difference is that the router in the broadcast subnet sends the link status update to the multicast addresses AllSPFRouter and AllDRouter. In the NBMA network, the link status update MUST be copied, then, they are sent to each neighbor router.
　　
4 Conclusion
OSPFv3, as an important internal gateway protocol in the IPv6 era, greatly enhances the flexibility and stability of the network. This article starts from multiple subnet technologies included in the Internet, this paper analyzes and proposes the implementation scheme of Dynamic Routing Protocol Based on NBMA link type OSPFv3. The Implementation Solutions for traditional point-to-point and broadcast link types are relatively mature, while the implementation of OSPFv3 is basically the same as that of NBMA links on point-to-point and broadcast link types, there is only a difference between DR/BDR election and neighbor/interface state machines in the Pan-flood mechanism. With the advent of the IPv6 era, how to efficiently run OSPFv3 dynamic routing protocols on a new generation of networks that are organically integrated with multiple link status types will be our goal Article entry: csh responsible editor: csh