A comprehensive explanation of CEF Routing Technology

The CEF Routing Technology in China has developed very rapidly. Many people may not understand the application principle and practical application of the CEF routing technology. After reading this article, you must have gained a lot, I hope this article will teach you more things. The basic role of a traditional router is routing computing and packet forwarding. Generally, based on the shared storage architecture, the shared bus is controlled by a centralized CPU, that is, a single CPU or multiple CPUs, which are connected to a router cluster, connect to multiple interface cards. The interface card contains simple queues and other structures to communicate with the CPU.

Data Packet forwarding is achieved through the shared bus. With the rapid development of the Internet and the emergence of a large number of new service requirements, it puts forward higher requirements on the network routing and switching performance. It is necessary to increase the packet forwarding rate and system performance at the same time, the design architecture of traditional routers and switches must be improved, and some new design schemes must be added to improve the system performance. CEF routing technology CiscoExpressForwarding, Cisco Express exchange) is a new route exchange solution launched by Cisco, which has good switching performance, enhanced switching architecture and extremely high packet forwarding rate. GSR1200 series Gigabit exchange routers using CEF routing technology have qualitative changes in architecture, routing mode, and interface card performance, especially suitable for the core layer of ISP networks with large business volumes, it is also widely used as the backbone of High-Speed enterprise networks. This article introduces and analyzes the express exchange technology of CEF routing technology.

I. a brief review of vro switching algorithms

1. Process exchange

The original Cisco router used centralized CPU packet switching. All packets were transferred to the CPU through the shared bus. The route table was used to search for the packets, and the CRC was recalculated. Then, the packets were uploaded to the appropriate line card through the shared bus.

2. Quick exchange

An IP packet that arrives at a specific destination address usually results in a packet flow. That is, if the packet is switched to a specific destination, the other packet may arrive soon. By constructing a high-speed cache for the nearest exchange destination, you can reduce the number of times the package searches for the same destination in the whole routing table. This "one route, then switch" method is called fast exchange, fast Switching greatly improves the packet forwarding rate of the router, and thus becomes the default switching mechanism on the Cisco router platform. However, note that changes to the IP route table must cause high-speed caching to be ineffective. in an environment with constantly changing routing conditions, the advantage of high-speed routing caching will be greatly limited.

3. Autonomous exchange

Autonomous switching is characterized by the absence of some switching functions from the CPU. In effect, the route high-speed cache function is moved from the CPU to the secondary exchange processor. The receiving packet on the line card first searches for the local route high-speed cache target in the exchange processor, if the search fails, the CPU is interrupted and the route table is queried. Here, Cisco renamed the CPU of the periodic computing route as the route processor and the auxiliary exchange processor as the switch processor. The Cisco7000 series routers implement autonomous switching to further improve throughput and other performance.

4. distributed switching

With the introduction of VIPVersatileInterfaceProcessor, universal interface processor) card, the vro switching system gradually develops to the peer-to-peer multi-processor structure. Each VIP card contains a RISC processor, which maintains the latest copy of the fast exchange cache generated by the route exchange processor, and can independently implement the routing switching function, two types of switching are completed at high speed-local VIP switching and VIP switching.

5. CEF Routing Technology

As mentioned above, the fast-switching high-speed cache mechanism often has network topology changes, route changes, route fluctuations, and so on in high-speed dynamic routing selection environments such as the Internet, the change in routes leads to high-speed cache invalidation, while rebuilding the high-speed cache is the process of "process switching".) the computing overhead is huge. At the same time, with the rapid development of the Internet and its services, WEB-based applications and interactive services increase the number of communications and short Real-time Data Streams. The fast-exchanged cache content is constantly changing, increasing the burden of rebuilding the cache, as a result, the performance of the router is reduced. The CEF routing technology is designed to address the above shortcomings.

Ii. Basic principles of CEF Routing

1. CEF parts

The CEF routing technology is an advanced layer-3 switching technology designed for high-performance and high-scalability layer-3 IP backbone network switching. To optimize the routing lookup mechanism for packet forwarding, CEF routing technology defines two main components: the forwarding information library ForwardingInformationbase and the AdjacencyTable ). Forwarding information library FIB) is a query table used by the router to determine the target exchange. The FIB entries have a one-to-one relationship with the IP route table entries, that is, FIB is an image of the route information contained in the IP route table. Because FIB contains all required route information, you do not need to maintain the route cache. When the network topology or route changes, the IP route table is updated, and the FIB content changes accordingly. The CEF routing technology uses the adjacent table to provide the information required for data packet Rewriting on the MAC layer. Each item in FIB points to a next hop relay segment in the adjacent table. If adjacent nodes can forward each other through the data link layer, these nodes are listed in the adjacent table.

Once the system discovers the adjacent relationship, it writes it to the adjacent table, and the adjacent sequence is generated at any time. Each time an adjacent entry is generated, a link layer header information is calculated for the adjacent node in advance, the link layer header information is stored in the adjacent table. When the route is determined, it points to the next network segment and the corresponding adjacent entries. Then, it is used for encapsulation when the packet is switched to CEF. To view information about the adjacent table, run the command showadjacency/showadjacencydetail of CiscoIOS. When we view the information about the adjacent table, we can find two main types of adjacent data: Hostadjacency and PointtoPoint. Generally, the Hostadjacency type is displayed as an IP address, which indicates the next hop IP address of the adjacent node. The PointtoPoint type is displayed as "point2point", indicating that this is a point-to-point circuit. There are other special types, such as Nulladjacency and Gleanadjacency.

2. CEF Operation Mode

There are two CEF modes: centralized and distributed. Centralized allows a routing processing module to run express exchange, that is, FIB and the adjacent table reside in the routing processing module. When the line card is unavailable or does not have the function of distributing the CEF routing technology exchange, you can use the centralized CEF routing technology exchange mode. Distributed is generally recorded as CEF) allows multiple line cards of the router to run the express exchange function separately, provided that the line is a VIP line card or GSR line card. The Central Route Processor performs system management, route selection, and forwarding table computing, and distributes the CEF table to a single line card. Each line card maintains the same copy of one FIB and the adjacent table. The line is stuck between the port adapters for fast forwarding, so that the switch operation does not require the involvement of the Routing Switching Module. The CEF routing technology adopts an internal process communication mechanism to ensure the synchronization of FIB and the adjacent table between the route processor and the interface card.

The Cisco 12000 series routers only run the dCEF mode and are switched by the line card. In other routers, you can use interfaces of different types in the same vro. If an interface card that does not support the CEF Routing Technology receives a packet, the data packet is forwarded to the Route Processor for processing, or the data packet is forwarded to the next network segment for processing. The CEF routing technology is globally activated on the vro, but CEF can be enabled/disabled on each interface or the VIP's baseboard. The CEF and quick switch modes can also run simultaneously, this is not recommended because it occupies a large amount of system maintenance resources.