Main Structure and technical analysis of LAN switches

Source: Internet
Author: User

There are four main types of LAN switches: memory-type structure, cross-bus structure, hybrid cross-bus structure, and ring-type bus structure. The superior performance of LAN switches comes from its unique internal technical structure. Different switching modes or different switching types are also closely related to the internal structure of LAN switches. Therefore, understanding the internal structure of a LAN switch is equivalent to understanding the technical characteristics and working principles of the LAN switch. Currently, LAN switches use the following internal technical structures.

1. Shared Memory Structure

This structure depends on the high-performance connection of the full port provided by the central LAN switch engine, and the core engine checks each input packet to determine the connection route. This method requires a lot of memory bandwidth and high management costs. Especially with the increase of LAN switch ports, the memory capacity is required to be larger and the speed is faster, and the price of central memory becomes very high, this makes the LAN switch memory a major bottleneck for performance implementation.

2. Cross-bus structure

A cross-bus structure can establish direct point-to-point connections between ports. This structure provides good performance for simple single-point (Unicast) information transmission, but is not suitable for point-to-point Broadcast Transmission. In the actual network application environment, the broadcast and multicast transmission modes are very common, so this standard cross-bus mode will bring about some transmission problems. For example, when port A transmits data to Port D, port B and port C can only wait. When port A broadcasts messages to all ports, it may cause queuing of the target port. This will consume a large amount of system bandwidth, thus affecting the transmission performance of LAN switches. N × (N + 1) Cross bus is required to connect N ports. Therefore, the implementation cost will also increase sharply as the number of ports increases.

3. Hybrid cross-bus structure

In view of the defects of the standard cross bus, a hybrid cross bus implementation method is proposed. The design idea of this method is to divide the integrated cross Bus Matrix into small cross matrices, and connect them through a high-performance bus in the middle. This structure reduces the number of Cross buses, reduces costs, and reduces bus contention. However, the bus connected to the cross matrix has become a new performance bottleneck.

4. Annular bus structure

This structure supports a maximum of four switching engines in a ring, supports Switching Matrix interconnection at different speeds, and intercommunication between the ring and the ring through the switching engine. Because the ring structure is used, it is easy to aggregate bandwidth. When the number of ports increases, the bandwidth increases accordingly. Different from the preceding structures, this structure has an independent control bus used to collect bus status, process routes, control traffic, and clean up the data bus. In addition, the management module can be added to the ring bus to provide complete SNMP management features. You can also select the layer-3 switching function as needed. The biggest advantage of this structure is its strong scalability, low implementation cost, and effectively avoids the bus bottleneck caused by system expansion.

Main technologies of LAN switches

Because a LAN switch uses a virtual line exchange method, it is technically possible to use different bandwidths between input and output ports, or without a transmission bottleneck, the high-speed transmission of data between ports greatly improves the data transmission of network information points and optimizes the network system. The main difference between LAN switches and hubs in hardware is that there are more Backplane Buses and exchange engines. This shows that LAN switches have a high technical level. Therefore, to fully understand LAN switches, you must understand the main technical features of LAN switches. The following describes the main technologies used in LAN switches.

1. Programmable ASIC (Specific Purpose IC)

This is a dedicated IC chip dedicated for optimizing Layer 2 switching processing. It is also the core integration technology of the current networking solution. It can integrate multiple functions on the same chip, it has the advantages of simple design, high reliability, low power consumption, higher performance and lower cost. The programmable ASIC chip widely used in LAN switches is an ASIC chip that can be edited by manufacturers or even users based on application needs, it is one of the important application technologies in LAN Switch Applications.

2. Distributed Pipeline

With the distributed pipeline, multiple distributed forwarding engines can quickly and independently transmit their respective data packets. In a single pipeline, multiple ASIC chips can simultaneously process multiple frames. This concurrency and pipeline can increase the forwarding performance to a new level. Provides line rate performance for on-demand (Unicast), Broadcast (Broadcast), and Multicast (Multicast) on all ports. Therefore, the adoption of distributed pipelines is an important reason for the increase in the exchange speed of LAN switches.

3. dynamically Scalable Memory

For advanced LAN switching products, high performance and high quality features are often built on an intelligent storage system. The dynamic Scalable Memory technology enables LAN switches to dynamically expand the memory capacity based on the needs of data streams during operation. To this end, part of the memory has been directly associated with the forwarding engine in the layer-3 LAN switch mode, so that it can add more interface modules. In this way, including their respective forwarding engines, the memory will be extended accordingly. At the same time, you can also dynamically construct the cache through pipeline ASIC processing to increase memory usage. This also prevents packet loss when the system processes large burst data streams.

4. Advanced queue mechanism

In fact, no matter how good the performance and quality of network devices are, no one will be affected by the Data congestion on the connected network segment. The traditional method is that the traffic through a port must be stored in the cache of only one output queue. No matter how high its priority is, it must also be handled in the first-in-first-out mode. When the queue is full, any excess parts will be discarded. When the queue grows, the latency will also increase. Obviously, the traditional queue mechanism makes it very difficult to run real-time transaction processing and multimedia applications. To this end, many network equipment vendors are developing Advanced Queuing technologies to provide different service levels on an Ethernet segment and control latency and jitter. The advanced queue mechanism can be that each port has a queue mechanism of different levels, which can better distinguish different traffic levels, so that the network system can better match with high-performance applications. Data packets such as multimedia and real-time data streams are put into high-priority queues. After the Weighted Fair queuing algorithm is used, data packets can be processed in high-priority queues more frequently, it does not ignore low-priority queues. In addition, traditional application users do not notice changes in response time and throughput, while those who use emergency applications can receive timely responses.

5. Automatic traffic classification

In network transmission, some data streams are more important than other data streams. The layer-3 LAN switch has adopted the automatic traffic classification technology to distinguish different types and levels of data traffic. Practice has proved that after the automatic traffic classification technology is used, the layer-3 LAN switch can instruct the data packet assembly line to differentiate user-specified data streams, thus achieving low-latency and high-priority transmission, it not only provides effective control and management channels for special data traffic, but also avoids network traffic congestion.

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.