In-depth discussion on the Origin and Development of Route Exchange Technology

Routing switches are quite common. So I have studied the origin and development of the routing switch technology. Here I will share it with you and hope it will be useful to you. First, let everyone understand the origin and development of the Routing Switching technology, and then introduce the application and trend of the L3 Routing Switching technology. It is necessary to establish an end-to-end connection in advance by the ATM network, and then use the "route cut" method to Route IP packets.

The RFC1953 In the IPSwitch method solves the problem of increasing the number of "multi-hop" requests. The software provides a "straight-through" Cut-through) to meet the business requirements of multiple IP addresses, it works with RFC1987 to form the basic of IPSwitch. In most cases, IPSwitch processes data packets in a CUT-THROUGH mode. That is, the first packet is checked, identified, and processed by the router, later, the same packet will be transmitted through the ATM switch and will not pass through the router.

Whether IPSwitch or MPOA, this IP data stream is transmitted in a virtual circuit. All IP packets are transmitted in a selected route. Different IP packets are routed through different routes. Only in IPSwitch mode, each ATM switch can independently process IP exchange to pass through IP data streams. However, MPOA must act in a uniform manner on all ATM switches. Therefore, an end-to-end SVC must be established before MPOA is implemented. In addition to the preceding two L3 Routing Switching technologies, the Routing Switching technology has also been developed in other fields. For example, Cisco's proprietary technology CEF Cisco fast forward) the Multi-Layer Routing Switching Technology MLS widely used by all third-tier switch manufacturers and the widely-promoted Multi-Protocol Label Switching MPLS based on the IETF standard.

Origin and Development of L3 route Exchange Technology

L2 Ethernet Switching Technology-Based Multi-layer switching originated from campus networks. Later, there were many applications in IDCs. Early Internet business traffic models comply with rules 20: 80, that is, 80% of the traffic is local, and 20% of the traffic is outbound. Later, the traffic model was reversed. 80% of the traffic came from outside the network segment, and only 20% of the internal communication was performed. As a result, more and more services are exchanged between different network segments. Each L2 CIDR block represents a broadcast domain. Due to network scalability constraints, you need to divide different VLANs for different CIDR blocks on the L2 Ethernet switch. We know that in a multi-VLAN environment, l2 switches are used inside VLANs, and L3 routes are used between VLANs. Therefore, L3 data packets must be forwarded between VLANs through routers. In the traditional routing mode, the processing performance of IP packets is lower than that of Ethernet frames. With the increase of cross-network services, the conflict with traditional L3 packet forwarding performance is growing. These factors promote the generation of L3 Routing Switching technology.

The difference between L3 switches and traditional routers is that, in addition to CPU-based Softswitch, L3 switches use ASIC. The main difference between L3 switches is the forwarding mechanism. L3 switching meets the layer-3 routing requirements while meeting the line rate forwarding requirements for data. Only by improving the throughput can the bottleneck problem on the layer-3 be completely solved. L3 switching combines the advantages of the Routing Switching technology, and its high performance is achieved based on the "one-time routing, multiple-times switching" mechanism. During the L3 switching process, the L3 switch monitors the first Ethernet frame that enters the router interface based on the destination address of the Ethernet frame as the MAC address of the router interface), because these data frames must be cross-network traffic. Then, the L3 switch needs to strip the Ethernet frame and read the L3 information. By looking for the FIB route forwarding table, the packet is output to the corresponding interface. At the same time, the L3 switch classifies the Flow of data packets that subsequently enter the switch based on the IP address source/destination address of the data packet, cache each stream and the output port of the stream. Subsequent data packets in the stream do not need to search the route table for the destination address of the data packet for L3 forwarding each time, you only need to replace the destination MAC address of the Ethernet frame that encapsulates the L3 data packet. This achieves one route and multiple exchanges.

Because multiple data streams can be created between two endpoints on the Internet at the same time, with the same source/destination IP address, if the device can classify the stream based on more fields in the IP data packet, for example, based on the protocol type and the TCP/UDP port number, you can classify each Session. The layer-4 switching can be called L4 switching. L4 switching is required if the policy requires refined Traffic Control Based on the application or traffic statistics based on the application. The switching principle is the same as that of L3 switching, But it consumes more device resources, CPU and memory ).

Application and trend of L3 Switching

With the increasing maturity of the routing exchange technology, layer-3 Switch networking technology is increasingly used for carriers, networks, and campus networks. It brings obvious advantages. First, low network construction costs and flexible networking. The routing and forwarding performances of the L3 vswitch are no less than those of the gibit router, but they have higher port density. Each port of the L3 switch can be flexibly configured as a switching port or a routing port. It also provides a POS interface for interconnection between devices. IGP can be run through the POS port to provide the routing function, implement Internet access services; provide Ethernet passthrough services for VLANTrunk through GE port interconnection configuration. Traditional routers end up L2 and cannot provide L2 passthrough services.

Second, to a certain extent, it meets multiple business needs. Currently, data services can be divided into leased line services and Internet access services. The Routing Switching technology can provide two services on one device. In Internet businesses, for broadband residential users of Ethernet users, the CPE side is connected to the L3 switch of the ISP metropolitan area network through a router or a FE/GE Interface through bare optical fiber). For ADSL users, after being aggregated to DSLAM, it is terminated to BRAS through FE/GE uplink over Metro Ethernet. This is based on the routing function of the L3 switch to achieve Internet route accessibility.

Currently, L3 switches with excellent performance support the same routing protocols and policies as traditional routers, for example, PolicyRouting for BGP Route entry support, IGP convergence performance, and routing policy implementation, and CAR for VLANID or port) NAT, ACL, EtherType field filtering, and broadcast packet filtering in terms of Network Security) and IPPrecedence bit ing in QoS for 802.1p priority queues, and between 802.1p and ToS bytes ), even MPLS applications provide better solutions. For leased line services, the L3 switch network has inherent superiority in providing Ethernet passthrough services. Especially for point-to-multi-point services, VLAN stacked applications, and port binding, it provides good performance.

Of course, no device, technology, or networking solution is omnipotent. The fault of L3 switch networking is that the network loop protection is restored through the Spanning Tree Protocol) the impact on network scalability, ease of management and maintenance, and the ability to support IGP routing, VLANTrunk, and MPLS switching in the L3 switch Interconnect Link are insufficient. In the future development trend, the most perfect situation is to gradually and completely replace traditional routers to achieve all the functions supported by routers, especially the support capability of MPLS, such as MPLSTE, L3MPLSVPN, and L2MPLSVPN. By then, with the further improvement of MPLS technologies, applications and standards, the Metro Ethernet built with L3 Ethernet switches will be able to provide real full-service services, such as known Ethernet passthrough to point-to-point and point-to-point), Internet access, and even use L2MPLS technology to implement AToM.