|
Switching Technology
CrossBar (that is, CrossPoint) is known as cross-switch matrix or crossbar switching matrix, which is recognized by the industry used to build large-capacity system of choice for the exchange of network structure. Said the three are to some extent, the exchange of the structure of the shared bandwidth, and Crossbar switching network structure completely broke through such restrictions, in the exchange network without bandwidth bottleneck, not because of insufficient bandwidth resources arising from obstruction.
Crossbar switching fabric
Crossbar switching network in the data plane did not Any bottlenecks. This is because the introduction of the exchange matrix Crossbar this new exchange, eliminating the sharing of bandwidth exchange, in the way data exchange is a revolutionary change. Crossbar switching network expansion capacity is very strong, the exchange can do a lot of capacity, the basic conditions are not the hardware, the current single-chip exchange capacity between the 256G-700G, multi-chip T-class and even to build the capacity of several large-scale T exchange network, sufficient to meet current and future years, the network of exchange capacity requirements, and hardware integration with the advances in technology, single-chip support Crossbar capacity may be even greater.
The performance of its advantages in the following areas:
1. First of all, CrossBar the realization of relatively simple
Share exchange structure switch fabric line cards to simplify the physical connection for point-to-point connection, to achieve more convenient and thus more easily guarantee the stability of large-capacity switch.
Secondly, CrossBar internal non-blocking. A Crossbar as shown in diagram, as long as at the same time closing a number of cross-node (CrossPoint), a number of different ports can transfer data at the same time. In this sense, we believe that all of the Crossbar is a non-blocking in-house, because it can support all the ports at the same time wire-speed data exchange.
Crossbar technology in support of the three switches, the general use of two types of three-tier exchange chips: one is a Gigabit, Fast port switch chip; a class is only the internal high-speed interface (often more than the rate of 10G ) The Crossbar chips, used in the interconnection between various modules.
At present, Cisco, Extreme, Foundry Networks, Harbor Networks, such as manufacturers have introduced Crossbar-based core switching products, but the market is that there is a lot of internal Gigabit port is still used for the Internet products, mainly on the performance requirements for the general, the price comparison sensitive user groups.
2. Wire-speed hardware-based access control
With the increase in the number of users, users need to MAC address, IP address, TCP / UDP port number and other information to control, in order to achieve a strictly limited access to LAN resources, but also limit the use of this feature LAN users of network equipment itself visit. Especially in the last in 2012 and the threat of invasion and occupation of network resources, network virus has greatly affected the stability of the three switches, control through the above-mentioned functions, the effective control of the user business, but also to effectively protect their own safety and reliability. At this time it developed into a three-switch "intelligent multi-layer switch," is mainly with the above control functions. This kind of switch is the exchange of chips support a deal with the contents of more than 64 bytes, so one can directly deal with the Ethernet MAC address of the first frame, IP header, TCP / UDP Baotou, in order to achieve a 2 ~ 4 layers of hardware-based wire-speed access control. There are some intelligent control functions such as 802.1X authentication protocol support, through the RADIUS protocol, such as the expansion of AAA to the user more stringent control.
3. End-to-end QoS technologies
ASIC technologies enable the rapid development of low-end equipment with powerful QoS capability to become possible, the network QoS guarantee gradually began to focus on the transition to the end-to-end guarantee. Now, the network edge devices are based on port, MAC address, VLAN Information, IP address or even high-level information to identify the type of application for priority packets marked with tags (such as modification or IPDiffServ domain IEEE802.1p), the core devices do not need to Application for recognition, just under the IPDiffServ and IEEE802.1p exchange, and provide quality of service can be. QoS features such intelligence is based on chips to deal with more than 64-byte depth developed.
4. A richer protocol support
Now intelligent multi-layer switch to support the ordinary two-story addition to the agreement, but also to support the BGP, IS-IS routing protocols, such as rich, as well as PIM Multicast Routing Protocol, VRRP redundancy backup agreement, etc. 2 to 4 layer agreement. Extreme, Harbor and other manufacturers and even the introduction of the Ethernet switch to support carrier-grade 50ms ring network technology related agreements, making three switches can be applied to more complex and demanding environment.
In addition to the above number of technical changes, the three-port switch also supports the three-tier clustering technology, NTP (network clock Agreement) and other functions, not here to elaborate. With the three-tier exchange and related technology development, the three-switch products have been a lot of broken down, according to different functions can be divided into three cassette Fast switches, the entire three-cassette Gigabit switches, rack-mounted modules of the three switches, in order to meet future needs of different networks.
|