Processing method and device in multi-core system

Abstract

The invention provides a processing method and a system in a multi-core system. A multi-core comprises a first core and at least one second core. The method comprises the following steps of binding a control progress on the first core; binding data progresses on the at least one second core, wherein each data progress corresponds to one second core; establishing a private pile and a sharing pile for the control progress and the data progresses, wherein each core is an authority core of the sharing pile or the private pile of the corresponding progress, the private pile is only visible to the control progress or the data progress bound on the authority core, and the sharing pile is visible to the control progress or the data progresses bound on all the cores; and distributing the private pile of the control progress to be used for storing private information, and distributing the sharing pile of the control progress to be used for storing information related to the data progresses. According to the method disclosed by the embedment of the invention, synchronous expenses of multi-core configuration can be avoided, and the processing performance of the multi-core system is improved.

Description

technical field [0001] The present invention relates to the technical field of the Internet, in particular to a processing method and device in a multi-core system. Background technique [0002] In the process of message processing, when the control plane configures the data plane, in order to ensure the consistency of configuration information, a synchronization mechanism is required. Especially in a system based on multi-core CPU, as the number of cores increases, the Synchronization overhead can grow rapidly and severely impact system performance. [0003] The existing technology adopts a multi-core message processing device that mixes the control plane and the data plane. The control plane and the data plane run on the same core, and the configuration information and statistical information between the cores are When a document needs to find or modify shared information, it needs to adopt a synchronization mechanism, for example, a spin lock, which locks the information...

AI Technical Summary

Problems solved by technology

[0004] The problem existing in the existing technology is that, on the one hand, due to the need to adopt a synchronization mechanism, as the number of cores increases, the synchronization overhead between cores will increase rapidly, so that the system performance cannot increase linearly with the increase in the number of system cores, and when the number of cores When the number increases to a certain extent, the performance decreases instead. For example, in the LVS (Linux Virtual Server) system of the prior art, the forwarding performance of the system becomes a log-shaped curve as the number of cores increases.

On the other hand, due to the frequent update and modification of configuration information and statistical information on each core, a large number of cache synchronization actions between cores will be triggered, which will affect the performance of the system and cause serious cache misses.

Images