Application viewpoint: development status of core router technology

With the development of the routing industry, the router technology is constantly updated and upgraded. Here we mainly analyze the development history and future development direction of the broadband core network technology. In recent years, the explosive growth of IP services in the world has caused a huge demand for network bandwidth. The IP data network bandwidth requirement doubles every six months, surpassing the famous CPU Moore's Law.

At present, the IP data volume of many international communication operators exceeds that of the voice service. It is expected that the IP Data Service of the worldwide communication network will surpass that of the voice service in the next few years. In the future, voice services will even become only affiliated services, and IP addresses will become an undisputed unified platform for "three-in-one" in the future. As we all know, the existing human communication network is a dedicated network built to deliver a specific business, so there are also a variety of network router technologies, it mainly includes Ethernet for data communication, ATM for telecommunication networks, SDH, and DWDM. In the future evolution of a Unified Communication Network, we must try to protect existing network investment as much as possible, and there must be innovation, the following is a brief introduction to the existing and future core router technologies.

IP over ATM

ATM has been regarded as a perfect technology used to unify the entire communication network. In the future, all voice, data, video, and other services will be transmitted through ATM. International, especially the Telecommunications Standardization Organization, has studied the router technology for many years and has been applied in practice. However, as expected, the ATM did not achieve the expected goal. At the same time, the development speed of IP is greatly unexpected, but on the one hand there was no independent IP backbone network router technology from beginning to end a few years ago, on the other hand, during the rapid development of the IP address, it does have some defects, such as low QoS. Therefore, IP over ATM (IPOA) technology is first generated in the broadband IP backbone network.

The basic principle of IP over ATM is to encapsulate all IP packets in the ATM layer as ATM cells and transmit them in the channel in the form of ATM cells. When a vswitch in the network receives an IP packet, it first processes the route address based on the IP address of the IP packet and forwards it by route. Then, create virtual circuit VC on the ATM network based on the calculated route ). In the future, IP data packets will be transmitted Through the Cut-Through mode in the virtual circuit VC and downloaded Through the router, which effectively solves the bottleneck problem of the IP router, the forwarding speed of the IP packet is increased to the switching speed. The difficulty of IP over ATM technology is how to combine the non-connection of IP addresses with the connection-oriented router technology of ATM. There are many IP over ATM technologies, which can be divided into two models: Overlap Model and integration model.

1) Overlapping Model

Overlapping models are implemented in the following ways: IPOA of IETF, CIPOAC1assic IP over ATM), lane lan simulation of ATM Forum, and MPOAMulti-Protocol over ATM. The main idea of overlapping technology is that IP routing is still implemented by IP router technology. IP Address Resolution Protocol ARP is required to map MAC addresses to ATM addresses or IP addresses to ATM addresses. The host does not need a traditional router. Any host or edge device with the MPOA function can be directly connected to another device through ATM exchange, packet Switching is completed by the edge device, that is, layer-3 packet switching. The signaling standards of this technology are well-developed and compatible with standard ATM networks and services, adopting the signaling standards of ATM Forum/ITU-T. However, this technology only supports multicast services within the logical subnet. multicast between subnets must use the traditional router technology, which is less efficient for broadcast and multi-service.

2) Integration Model

The implementation technologies of the integrated model mainly include Ipsilon's proposed IP Swtich technology), Cisco's label exchange Tag Swtich technology and the MPLS Router technology recommended by IETF. The main idea of the integration model is to regard the ATM layer as the Peer layer of the IP layer, and combine the router technology of the IP layer with the switching function of the GN layer to enable the IP network to obtain the routing function of the ATM, the ATM endpoint only needs to be identified by an IP address, so the Address Resolution Protocol is not required. Due to the high overhead of IP over ATM, which is as high as 24%, and other reasons such as cost of network equipment and network bandwidth expansion, IP over ATM is generally only used on the edge of the network.

IP over SDH

The basic idea of the POS defined by IETF is to map IP datagram directly to SDH frames through the Point-to-Point Protocol (PPP), eliminating the complicated ATM layer in the middle, which can greatly save network investment. The specific method is to encapsulate the IP datagram into PPP, then use the High-Level Data Link Control (HDLC) to form a frame, and then map the bytes to the virtual container (VC) encapsulation, add the corresponding SDH overhead into the STM-N frame. IP over SDH is distributed hierarchically in the OSI (Open System Interconnection) model. In this scheme, the PPP protocol provides multiple protocol encapsulation, error control, and link initialization control, while the HDLC frame format is responsible for synchronizing the IP data frames in the PPP encapsulation on the transmission link. IP addresses appear in the form of packets on the OSI Layer 3; PPP appears in the form of frames on the OSI Layer 2; SDH appears in the form of frames on the OSI 1st ~ Layer 3.

PPP is short for the Point-to-Point Protocol. The header has only two bytes and has no address information. It is non-connected. This protocol can cut too long IP packets short (the IP packet length is unstable) into PPP frames to meet the requirements for SDH ing to SDH frames, it provides the features of multi-protocol encapsulation, error control, and link initialization control.

The main function of HDLC is to differentiate IP data packets transmitted over the synchronous transmission network using PPP encapsulation. This distinction is achieved through Byte Stuffing. Each HDLC frame starts with the Byte mark 0x7e and ends with 0x7e. At the transmitting end, HDLC frames are monitored to indicate the sequence and fill sequence. If the sequence occurs in the information field of HDLC frames, it is changed to 0x7d and 0x5e; on the contrary, in the fill sequence, change 0x7d to 0x7d 0x5e. At the receiving end, the filled information is discarded with only the original information fields. During idle time, when no datagram is transmitted, the HDLC mark is transmitted as a padding between frames.