The third layer switching first heard the word can not help but have some of the obscure, the ensuing exchange of the fourth floor, seventh floor exchange concept is the tough job faced by people. In fact, strictly speaking, the exchange means that the source and destination address the connection between the second layer of any of the above techniques can not be said to be switching technology. The term load balancing has been largely replaced the term fourth layer switching, just as the application of cognitive largely replaced the term as the seventh layer switching. However, I am afraid the word will always be the third layer exchange so called go. Of course, the argument is just saying, the key is to recognize that these technologies to improve network performance benefits, so this is still in use, "the first layers of exchange" that it is called. The third layer switching The third layer switching technology called IP switching technology, high-speed routing technology. This is a third-layer protocol used to strengthen the information exchange functions of the mechanism of the second layer. Most of today's enterprise networks have become the implementation of TCP / IP protocol within the network of Web technology, user data across the local network is often sent between the Internet and therefore, the router often overwhelmed. One way is to install more powerful super router, however, this overhead is too large, if it is building the exchange network, this investment is clearly unreasonable. Layer 3 switching goals, as long as the source and destination addresses between the second layer of a more direct path, there is no need to forward packets through the router. The third layer switching with Layer 3 routing protocols to determine the transmission path, this path can only be used once, can also be stored for later use. After the data packets through a virtual circuit to bypass the router quickly send. Layer 3 switching is currently the main technologies are: Ipsilon IP switching: IP switching technology, the company initiated by the Ipsilon, that identify the packet stream, as far as the second layer of exchange, to bypass the router, to improve network performance. Ipsilon improved ATM switch, deleting the controller software, coupled with an IP switch controller, and the ATM switch communication. The technology is applicable to in-house LAN and campus networks. Cisco Label Switching: to the packet label, the label read in the switching nodes to determine the packet transmission path. The technology is applicable to large networks and the Internet. 3Com Fast IP: focus on data policy management, the principle of priority and quality of service. Fast IP protocol to ensure real-time audio or video data streams to get the required bandwidth. Fast IP support for other protocols (such as IPX), can run outside in addition to other ATM switching environment. Clients need to set priorities for the software. IBM ARIS (Aggregate Route-based IP Switching): Cisco's tag switching technology with similar packages on the attached tag, to cross the switching network. ARIS is generally used for ATM networks, can also be extended to other switching technologies. Edge device is the entrance into the ATM switching environment, with the third layer routing maps to the second layer virtual circuit routing table. ATM network allows the same end of two or more computers to send data through a virtual circuit, thereby reducing network traffic. MPOA (MultiProtocol Over ATM): ATM Forum proposed a specification. The source of client requests, routing server performs routing calculations give the best transmission path. Then, create an SVC, you can cross subnet boundaries, do not do routing. Currently Cisco, 3Com, Nortel, Lucent, Cabletron, Foundry and Extreme are more mature companies such as Layer 3 switching products and modules available. Here to 3Com's technology, for example, to illustrate the evolution of the third layer switching technology. The first generation switch is a discrete electronic components and primitive-style software framework of the mixture. Software functions running on a fixed memory of the processor, the protocol functions as management support and improved software features are also increasing. When the user's daily business is more dependent on the network, network traffic increases, network equipment has become a bottleneck. Although the processor and memory to become faster and efficient, but still keep up with increased traffic levels. The first step is to simplify the problem-solving network layer: replace the router with a switch to reduce the packet processing overhead and significantly increase transaction speed. 3Com introduced to optimize the second floor dedicated to processing application specific integrated circuit (ASIC), the performance by 10 times, and lower overall system cost. Flexible Intelligent Routing Engine (FIRE) announced the advent of third-generation switching technology. This generation is not just based on the progress of the second generation, but also for the third layer routing, multicast (Multicast) and user-selectable policy (Policy) provides wire-speed performance and other aspects, the second layer and third layer inconsistent performance is no longer a. FIRE is the company's third-generation 3Com switches, the core of the third layer, which is an innovative integrated network interconnection architecture, providing a wide range of second and third tiers of functionality, but also in a variety of network Interface to provide wire-speed performance. The fourth layer switching End to end performance and service quality requirements of the load on all networked devices careful balance to ensure that the data between the client and server flows smoothly. The second layer and third layer switching products in the LAN and the Internet to solve the bandwidth and capacity to play a good role, but it might not be enough, need more performance, and this is the fourth layer switching arena. The fourth layer switching technology utilizes the third layer and fourth layer header information to identify the application data stream sessions, these include TCP / User Datagram Protocol (UDP) port number, mark the beginning and end of the application session "SYN / FIN "bit and IP source / destination address. Using this information, the fourth layer can make the switch to forward the session to the transport stream where the smart decision. For a variety of different application systems to support a large enterprise data centers, Internet service providers or content providers, exchanging the role of the fourth layer is especially important. Similarly, when a lot of copy on the server function, the fourth layer switching will also play no small role. Routers and switches in the third layer of a different packet forwarding does not understand what the first package which package in the post. The fourth layer switching technology to track and maintain each session from start to finish. Therefore, the fourth layer switch is a real "conversation switches." Router links or network nodes based on the availability and performance to make forwarding decisions, while the fourth layer switches are based on the session and application layer information to make forwarding decisions. As it reaches this point, so the user's request can be forwarded according to different rules to the "best" server. Therefore, the fourth layer switching technology is used to transfer data between multiple servers and achieve the ideal load balancing mechanism. Function with the fourth layer switch connected to the server to play with the "virtual IP" (VIP) front-end role. Each server and support a single or universal application of a server group are configured VIP address. The VIP address is sent out and the domain name registered on the system. When a service request is issued, the fourth layer switch by TCP to determine the beginning, to identify a session begins. It then uses sophisticated algorithms to determine the best server processing the request. Once such a decision, the switch will be a session with a specific IP address associated with, and use the real IP address of the server on the server instead of the VIP address. Each switch is to save a fourth layer with the selected server to match the source IP address and source TCP port associated with the connection table. Then the fourth layer switch connected to this server to forward requests. All subsequent packets between the client and server re-mapping and forward, until the session switch found so far. In the exchange with the fourth layer in the case, access can be connected with the real server to meet the rules established by the user, such as to have an equal number on each server access or depending on the allocation of server capacity to transport stream. It is generally single-function products can be load balancing 400-800 connections per second access. While the second layer and fourth layer has a function of a new generation of products (using a custom ASIC hardware-based load balancing) the connection speed is more than 10 million times per second access. In all this the key question is how to determine the transmission flow distributed to the most available server platform which, at present, making load balancing decisions using a variety of ways. According to the required load balancing granularity, the fourth layer switch can use a variety of methods to assign to the server application session. These include the right to seek access to the simple weighted minimum number of cycles, measuring round-trip delay and the server itself closed loop feedback, and so on. Closed loop feedback is the most advanced method, which uses available memory, I / O interrupts and CPU utilization and other system-specific information, which can drive adapter and switch automatically to obtain the fourth floor. The current closed loop feedback mechanism required to install software agents on each server. The fourth layer switches in the form and function with dedicated load balancer is entirely different. Traditional hardware-based load balancer is to optimize the speed of 45Mbps two-port device. The fourth layer switch is designed for high-speed Intranet application that supports 100Mbps or Gigabit interfaces. The fourth layer switching functions in addition to load balancing also supports other features such as application type and user ID based on the traffic control functions. Multi-level queuing technology, the fourth layer switch can be used to mark traffic as well as the traffic allocation priority. In addition, the fourth layer switches placed directly in front of the server, it understood the contents of the application session and user permissions, thus making it to prevent unauthorized access to the server an ideal platform. The fourth layer switching products User swarmed past the second and third tiers using switches, because these switches improve overall network throughput, making it far more than the old technology, throughput, do not know whether the fourth layer switch will see this phenomenon. Berkeley Networks company exponeNT e4 and Alteon Networks's ACEswith 180 two fourth layer switching products with outstanding performance and flexibility, will be better than the second and third floor switch to make more intelligent forwarding decisions. As the header code embedded in the query in the switch-specific integrated circuit (ASIC) to achieve the above functionality in almost will not cause any delay. The two companies can realize the switch 10M, 100M and Gigabit Ethernet capabilities, but Berkeley's switch is designed for enterprise applications, while the Alteon switch is used for Web or FTP servers with large institutions. Alteon's fourth layer switching technology through the server performance and health of the real-time monitoring, depending on the health of the server, the data flow will be visiting a cost-effective way to allocate to the appropriate server. Meanwhile, Alteon fourth layer switching technology with the Web cache redirection can specify to the remote Internet host to intercept HTTP traffic and redirect such traffic to the local cache server, thus greatly accelerating the access to the Internet speed, and saves a lot of valuable WAN bandwidth. And this for the users and providers of information is completely transparent to users and information providers do not need to do any settings.
No comments:
Post a Comment