The future development and application trend of th

  • Detail

The future development and application trend of layer 3 switch

Layer 3 switch is a switching technology that began to appear around 1997, initially to solve the problem of broadcast domain. After years of development, three-layer switching technology has become the main force in building a multi service integration network. At present, what changes are the application environment of layer 3 switches facing? What new technologies have emerged? What is its development trend? How about the future market demand? See experts' Interpretation -

production and development

Layer 3 switching (also known as multi-layer switching technology, or IP Switching Technology) is a switching technology that began to appear around 1997. It is proposed relative to the traditional switching concept. As we all know, the traditional switching technology operates in the second layer of OSI network reference model - data link layer, while the three-layer switching technology realizes the high-speed forwarding of data packets in the third layer of the network model. The emergence of three-layer switching technology is mainly to solve the broadcast domain problem in large-scale networks, through VSP? column=news&key=LAN target=_ Blank>lan divides a large switching network into several smaller broadcast domains, and three-layer switching technology is used between VLANs. The original three-layer switches often combined layer-2 forwarding and layer-3 switching in two units, and did not use a chip to complete the complete layer-3 switching function. Such switches are often rack type, such as 3Com's corebuilder 9000 and corebuilder 3500, Cisco's 5505 and 6509, Lucent's Cajun P550, etc. generally, there is a unit or module dedicated to processing layer-3 data

in traditional switches, the three-layer switching engine is often the bottleneck of the whole switch, which cannot achieve high-capacity line speed three-layer switching, and the bus structure will be adopted between modules. After the emergence of Gigabit Ethernet standard, some rack switches also use Gigabit ports to realize the interworking between modules. In 1998, Intel launched 550t and 550s stackable box type three-layer switches, with a backplane capacity of 2.1gbps, which can realize line speed switching of eight 100m ports. This was one of the earliest box switches in the market at that time, and its cost performance was relatively high. However, whether it was a box type three-layer switch or a rack type three-layer switch at that time, the main function was only to isolate the broadcast domain. The support of routing protocols was relatively simple. It only supported the dynamic routing protocols of small networks such as rip and OSPF. The routing between VLANs was also interworking by default, and there was no control function

with the change of network scale, Ethernet technology has moved from an office network to the local area of an office building and even the whole park. Before 1998, the park technology often adopted the earliest FDDI technology and ATM technology. This application change puts forward higher requirements for the performance of the layer 3 switch, the control ability of data forwarding and the routing interconnection ability between wide areas. At the same time, the reliability and availability requirements are also greatly enhanced. The layer 2 and layer 3 switching function has also developed to be completed by a single chip, and the switching capacity has also developed from the initial 5Gbps to the current level of hundreds of Gbps. Therefore, some key technologies have emerged, Such as crossbar technology, access control technology based on hardware wire speed, end-to-end QoS Technology, richer protocol support, etc. See the supporting document for details

changes in the application environment

even now that the layer 3 switching technology is quite mature, the development of the layer 3 switch has never stopped, mainly because the general metrological verification takes 10% or 20% of the maximum load of the equipment as the initial value of the verification. The application environment of the layer 3 switch is facing great changes

with the passage of time, the transmission speed of Ethernet gradually expanded from 10m to 100Mbps, 1Gbps and 10Gbps, and the price of Ethernet also fell rapidly with Moore's law and economies of scale. Nowadays, Ethernet has become the dominant network technology in local area network (LAN), and with the emergence of 10 Gigabit Ethernet, Ethernet is making great strides towards metropolitan area (man), which also drives the deeper reform of layer-3 switches. This pull is reflected in the following three aspects:

1 The construction of enterprise information infrastructure has brought huge space for Ethernet applications. Because Ethernet technology was originally designed for local area connection, it has natural advantages in supporting enterprise local area network connection. Its simplicity of construction, flexibility of expansion and continuous improvement of speed make it one of the preferred technologies for building enterprise network. Today, the vigorous promotion of China's information construction has brought great market opportunities to the development of ether

2. The construction of metro network has become a Xintiandi of Ethernet technology. At present, the architecture of metro network construction can be basically divided into mstp+ip and Ethernet mode. Ethernet technology plays an important role in the construction of metropolitan area because it is more suitable for the existing enterprise network connection and has the advantages of flexible and fast network construction. Ethernet based metro network is more suitable for data transmission and broadband growth. At the same time, the Ethernet structure is suitable for the coverage of large customers and business intensive areas, such as enterprise networks, campus networks, etc., so it has higher revenue expectations

3. The convergence service trend of Broadband provides space for Ethernet to move towards wide area. The integration of data services and traditional telecommunications services is the general trend. New operators look forward to a network that can provide integrated services and has better economy at the same time. IP based broadband Ethernet switching technology will gradually make this goal a reality. The 10G Ethernet technology for light has become a hot spot in the market, which makes the Ethernet technology go to the wide area, and the digital display semi-automatic impact experimental machine finally realizes the unified broadband network system from local area to wide area, realizes the support of integrated services, and forms the situation of Ethernet dominating the world

technology development trend

the rapid changes in the application environment faced by layer 3 switches have led to deeper technological changes in layer 3 switches. This change is mainly reflected in the following three aspects:

1 From the perspective of architecture, it flexibly supports the integration of various technologies.

the switching capacity of today's core switches has reached the level of hundreds of Gbps, which can meet the line speed forwarding of more than a dozen 10 Gigabit ports and hundreds of Gigabit ports. Therefore, performance is no longer a bottleneck. How to carry business well under the trend of network integration is a problem that all network equipment suppliers need to think deeply at the early stage of product design. The current network is the integration of routing and switching technology, wide area and local area, security, IDS and switch technology. It is mainly reflected in the direct expansion of firewall module, IDS module, 2.5g/155m POS, ATM, 2M and other router interface modules on the core switch. This convergence provides better flexibility for deploying various strategies in the network. For example, a separate firewall can only be deployed at the edge of the network to protect the security of the internal network at the outlet of the network, but more than 70% of the security problems often come from the interior, and it cannot effectively control the internal network. After being inserted into the switch as a module, the firewall can be flexibly deployed between any two VLANs, which greatly improves the flexibility of deployment. By expanding the router interface, users can save more investment and meet more flexible group needs

to achieve such integration, the following conditions need to be met in the switch structure:

● the system adopts a fully distributed crossbar architecture

that is, in addition to a core crossbar architecture on the switch board, each module connected to the switch board also adopts a crossbar chip to realize the standardized connection with the core switch board. The interface board module with this structure can adopt a very flexible structure. No matter what kind of chip structure is used on the S-type sensor module for the small force value of the sensor used by most manufacturers at present, as long as a crossbar chip is used with the switching board, it can realize perfect interoperability with each other. The advantages of this structure have two aspects: on the one hand, the crossbar chip on the interface board is directly connected to the two redundant switching boards, and the redundancy switching is directly completed through the crossbar chip on the interface board, without the participation of other parts, so it can achieve extremely fast recovery speed and achieve extremely high reliability; On the other hand, cossbar standardized any type of data on the interface board, making it possible to introduce other chip technologies into the switch. The disadvantage of adopting this structure is the relatively high cost

● both the software system and the hardware system have a fully distributed forwarding structure

under this architecture, the software of each interface module of the switch is relatively independent in operation, and the main control board software mainly plays the functions of table item management and synchronization. The second, third and fourth layer forwarding within each module does not need the participation of the switch board and the main control board, so it is possible to run the software with specific functions on the module

● introduce NP (network processor) technology into the switch

np chip is a chip between CPU and ASIC, and it is also a balance technology between CPU and ASIC. At the same time, it has the flexibility of CPU and the high performance of ASIC. NP technology is a major chip technology used by routers in the multi service convergence environment at present, which solves the problems of various protocol support and forwarding performance in the multi service environment. Now NP has been widely used in routers and firewalls. Under the two premises mentioned above, it is possible for the switch to introduce NP technology. By expanding the module containing NP chip, it realizes the support of specific functions such as expanding the wide area interface and firewall function module

nowadays, the emphasis on integration has been clearly reflected in various mainstream suppliers. Juniper's massive acquisition of security manufacturer NetScreen and Cisco's acquisition of IDS manufacturer riverhead are all products of this integration trend

2. Stronger multi service carrying capacity

stronger multi service carrying capacity is the foundation of the metropolitan area. With the trend of local and wide area integration, metropolitan areas are rising in scale. In the fields of e-government, education and scientific research, broadband metropolitan areas, VPN and other services are transferring from backbone to convergence. With the development of Ethernet switch chip technology and the improvement of the performance of convergence layer devices, especially the emergence of convergence feature core switches, MPLS VPN services originally mainly provided by backbone devices are gradually provided by convergence layer Ethernet switches. At first, the main reason for using backbone equipment to provide this service is that the performance of convergence layer equipment is insufficient, but now the performance of convergence layer Ethernet switch has exceeded the original backbone equipment; In terms of service provision, there are more convergence layer devices than backbone devices, which are closer to users and more convenient to provide services; From the perspective of network reliability, backbone equipment should develop towards function specialization, simplification and high performance due to its special position, while convergence layer equipment should take into account both performance and multi service support capabilities. This trend requires that core switches support perfect routing and switching characteristics, and the final device form is a device integrating routers and switches, so as to truly meet the needs of this market

3. Stronger and richer network monitoring and management capabilities

stronger and richer network monitoring and management capabilities are the basis for effective forwarding. Network management based on SNMP has become a consensus in the industry, and the operation of devices can be realized through RMON function

Copyright © 2011 JIN SHI