Content Recognition network is the management of network to the transport layer to the application layer. If a switch is able to unlock each layer of packets through each tier of packets, and identify the deepest information, then it has the content recognition function. In order to solve the problem of distinguishing application, dynamically allocating resources and billing of users, it is a good way to use network to identify devices to distribute traffic.
Some of the content-recognition devices that were implemented with software were initially presented, but did not achieve the desired results in the market. In the autumn of 2001, Networks+interop show the application of the latest silicon technology to implement the switch, the network experts have changed the long-term difficult to improve the 4~7 layer Exchange performance of the technical dilemma, so that content recognition network can be truly presented in front of users.
The shortcomings of software realization content recognition
Currently, there are three types of devices for content recognition by software, such as equipment on PC platform, layer 3rd switch with Universal CPU, and system based on network processor. The performance of these products will be acceptable to the user if they are simply done with simple traffic switching functions. However, this simple function can not let network management through the adjustment of networks, to obtain a profit value of the application administration. The point is that the information needed to complete these functions is buried deep inside the packet, and it only appears once when the network session is established. This requires that software-based content-recognition devices peek into each packet within each session, resulting in severe latency and performance degradation. Software-based content-recognition devices that rely on a general-purpose CPU or network processor cannot mobilize computing power in any near-real-time manner to complete a swap task, which can quickly become a bottleneck.
Advantages of extreme Hardware implementation
Extreme's application switching technology achieves comprehensive network functions, including line-speed Gigabit TCP session analysis, finalization, initiation, and even modification-all with hardware, eliminating complex software, general-purpose CPUs, and network processors.
Extreme's application switching technology (application Technology) is based on Pxsilicon. Pxsilicon is a unique, performance-excellence chipset. Compared with previous solutions, Pxsilicon performance is several orders of magnitude. Previous solutions have relied on complex software to work with a general-purpose CPU or network processor to accomplish the same load balancing task.
The transfer of network function from software to silicon wafer is a reasonable solution to this problem. This is not really a new idea, the late 90 from the software based routers, to the ASIC based on the 3rd layer of switch transfer can be proved. And, in any case, when the network technology is integrated into the silicon wafer, performance will be significantly improved, and the total cost of ownership will be significantly reduced. This solution has an intuitive result that service providers and enterprise users can freely set network applications and the rules required by the business without sacrificing the speed of gigabit performance.
The first platform to implement this new technology is the extreme SummitPx1 application switch. The SUMMITPX1 application switch supports a fully complementary layer 7th application switching capability, including the ability to parse Web requests, and the ability to connect to the most appropriate Web server with the requested content and server capabilities. It is also impressive that the SUMMITPX1 application switches maintain the gigabit performance of the line speed regardless of how much forwarding rules are set for the content.
The server selection algorithms for SUMMITPX1 applications include loops, weighted loops, minimum connections, and weighted minimum connections. It can also track client IP records, perform operations on client state settings (cookies), detect and track cookies, process cookies for server identification, and support persistent Secure Sockets Layer (SSL) session identification (IDS), and so on.