At present, the problem faced by network designers is that in the face of increasingly severe congestion load and impermanent congestion, the LAN must be effectively upgraded, and the investment required for the effective operation of the LAN must be maintained due to fiscal tightening. Faced with this situation, multi-layer exchange technology has emerged. For local area network (LAN) architecture, multi-layer switching technology combines the optimal characteristics of LAN switching technology and routing technology, and is a solution that improves performance-price ratio and has good scalability.
Switching and routing technology in LAN (local area network)
Switching technology in LAN emerged as the most effective solution to segment the shared LAN. LAN switching technology is in the second layer of the OSI network model, namely the data link layer. There is a certain number of physical ports in a LAN switch (usually between 8 and 128) for connecting to the LAN network segment. These ports obtain the relationship between the port and the destination address of the MAC by extracting the source MAC (Medium Access Control) address of each data packet sent to the switch.
The router works on the third layer of the OSI model, namely the network layer. Routers forward packets according to network layer address prefixes and routing allocation tables stored in router memory, which are constantly updated with the help of router protocols used to exchange reachable information between routers, so routers are more expensive and slower than switches.
Even so, routers have advantages: the current IP address mechanism is conducive to routing, but cannot be achieved by switches; in addition, routers also have IP filtering functions, which are safe, and switch-based local area networks are prone to broadcast blockage.
Multi-layer exchange technology
Multi-layer switching technology can support an integrated and complete solution for various LAN architectures, organically combining switching technology and routing technology intelligently.
1. Multi-layer switching equipment based on IP
The Ethernet multi-layer switching device has many switching ports, which can be logically regarded as a layer-second switching device with a layer-third forwarding function, and it uses high-speed interconnection with the third layer-third data forwarding module. During data communication, if the online site belongs to the same subnet, the second layer forwarding method will be adopted; otherwise, the third layer forwarding method will be adopted.
For example, suppose there are two sites using IP that communicate through a multi-layer switch and its Ethernet segments on both sides. When the sending site starts sending, the IP address of the destination station is known, but the MAC address of the destination IP is not yet known. At this time, it is necessary to determine through ARP (address resolution protocol). The sending station compares its IP address with the IP address of the destination station, and determines whether the destination station and the sending station belong to the same subnet through subnet blocking. If it belongs to the same subnet, the sender will arrive at the multi-layer switch with the MAC address of the destination station. It can be forwarded by simply searching the MAC destination station with the core of the second layer exchange. If the two sites are not on the same subnet, the sending site hopes to forward the packet through a "default gateway" (which can be regarded as a router), and the IP address of the gateway has been set in the system software. This IP address in a multi-layer switching device actually points to the third layer switching function block in the switch. Therefore, when the site broadcasts an ARP request to the IP address of the default gateway, the multi-layer switch will respond to the MAC address of the third layer forwarding module responsible for forwarding. Then, when the sending site starts to transmit an Ethernet packet with the destination MAC address, the second layer switching core sends the packet directly to the corresponding third layer forwarding module. At this time, the third layer forwarding module needs to broadcast an ARP request to obtain the MAC address of the final destination site and store it in the buffer. When each packet starts forwarding, the initial MAC destination address (pointing to the third layer forwarding module) has been replaced by a new MAC address pointing to the final destination site. The packet is then sent back to the second layer switching core, where the packet is directed to the correct output port by the MAC address table.
2. Multi-layer IP/IPX switch
As mentioned earlier, multi-layer switches use IP protocol (Internet protocol) to communicate at the second and third layers. However, Novell's network switching protocol (IPX) is more widely used in enterprise LAN than IP. Therefore, we need an ideal solution for IPX.
IPX and IP have different characteristics:
(1) The IP address is represented by 4 bytes, while the IPX address consists of 10 bytes.
(2) An IP subnet is defined as consisting of nodes with the same network or subnet number in a family, while an IPX subnet is defined as all sites within a broadcast domain.
That is to say, we can define the entire switching LAN as a separate broadcast domain, which can have multiple IP subnets, but can only have one IPX subnet. In this case, only the second layer forwarding technology for IPX is required in the multi-layer switch. However, in practice, unifying IP and IPX subnet divisions helps to maintain broadcast controllability. Since the broadcast protocol is closely related to IPX, the SAP protocol and RIP protocol are more likely to cause great broadcast congestion. One solution to this problem is to divide the network into a certain number of isolated broadcast domains, each of which contains an IP subnet and an IPX subnet. Only by adopting multi-layer switching technology can we define these broadcast domains logically rather than physically, which has been widely understood in recent years as a "Virtual LAN" (virtual subnet).
Obviously, if we have both IP and IPX in the LAN and choose to divide a switched LAN into multiple separate broadcast domains, we need a multi-layer switch to have a layer 3 forwarding technology that supports both IP and IPX, which is what we call a multi-layer IP/IPX switch.
Multi-layer switching routing protocol
We know that routers use RIP (Routing Information Protoco) and OSPF (Open Shortest Path First) to exchange subnet or inter-network information. The third layer forwarding function block of a multi-layer switch requires a conventional information exchange through the routing protocol standard like traditional routers.
In fact, in multilayer switches, there are two different ways to support routing protocols. One is to adopt a "self-inclusion" method, and the third layer forwarding function module in each multi-layer switch uses a routing protocol to achieve communication. Another method is to adopt the "routing server" method. A central functional block in the network adopts a routing protocol, which is responsible for one or more multi-layer switches, and applies some new and additional protocols to communicate, and updates the routing tables of multi-layer switches.
Performance of multi-layer switches
Local area network switches have reached a relatively ideal performance-price ratio. It is relatively easy to forward data packets from one network segment to another through hardware-based packet forwarding technology. This is because the forwarding policy is based on a simple MAC address (or RIF domain) lookup table and does not change the content of the packet.
On the contrary, the router is costly and the forwarding efficiency is not satisfactory. The reason is:
(1) The packet forwarding decision process in a router is much more complicated than that in a local area switch;
(2) Each data packet on the router must undergo security checks through the packet filter;
(3) Most or all packet forwarding processing of the router is implemented by software, making the performance price relatively low.
If a layer 3 forwarding function block is fully implemented by hardware, a multilayer switch will exhibit the same high performance regardless of whether it is switched on the second or third layer. However, if the third layer forwarding function is implemented in software, the forwarding data performance of the multi-layer switch will not be better than that of the router. So from the current perspective, there are still great differences between different multi-layer switch products, not at the same level.
Article entry: dnbm Editor: dnbm