Acceleration network card for cloud computing application layer

Abstract

The invention discloses an acceleration network card for a cloud computing application layer. The acceleration network card for the cloud computing application layer comprises a network card chip, a network interface and a host bus interface, wherein the network card chip is connected with the network interface and the host bus interface; a multi-core processing module is cured in the network card chip 1; a network protocol stack is cured in the multi-core processing module; the network interface is connected with a network; and the host bus interface is connected with a host. According to the acceleration network card, a multi-core technology is adopted, and the network protocol stack is cured in the network card chip, so that high-efficiency network transmission is realized; in the same situation, the concurrent access of more users and shorter transmission delay can be realized by a server, the deployment scale of the server can be reduced, and energy resources are saved; the supporting of application protocol processing is realized in hardware, so that the processing burden of the central processing unit (CPU) is uninstalled, and the efficiency of the system is improved; and the functions such as protocol stack timer management and message verification and computation can be realized in the hardware of the network card, and the access delay of the server can be reduced obviously.

Description

technical field [0001] The invention relates to network communication technology, in particular to an acceleration network card suitable for cloud computing application layer of high-speed network. Background technique [0002] In the fields of network security, network detection, and network testing, high-speed and efficient data transmission methods are key research topics in the industry. Under the trend of rapid development of high-speed network, the current network communication technology can no longer meet the demand. In the current solution, the ordinary network card receives the network data that needs to be served, and informs the processor through an interrupt, and the computer cpu processes through the tcp / ip protocol stack, submits the business data to the application layer business processing, and the cpu controls the core application layer business Then perform related business calculations. When new data needs to be sent to devices that need services, it wil...

AI Technical Summary

Problems solved by technology

In the current solution, the cpu is not good at tcp / ip business, resulting in a decline in its ability to handle core business

From this point of view, the current common network card has the following disadvantages: 1. Because the common network card lacks the ability to process network data, the cpu frequently responds to network packet interruptions, which consumes a lot of cpu time and leads to low efficiency; The copy from the kernel space to the application space occupies a large amount of memory bus bandwidth; 3. The timer management and cache management in the operating system protocol stack need to consume a lot of CPU resources; 4. On the mainstream multi-CPU multi-core server, it needs Traffic distribution among multiple computing threads makes it difficult to achieve effective load balancing

Images