Data distribution method and distributor

Abstract

The present invention relates to a data splitting method and splitter. The data splitting method comprises: a splitter parsing a received packet and determining the transport layer communications protocol to which said packet belongs; said splitter obtaining from said packet the identification information of the data stream, to which the packet belongs, corresponding to said determined transport layer communications protocol; according to the correlation between said transport layer communications protocol and a flow table, the splitter obtaining from the storage device the flow table corresponding to the transport layer communications protocol to which said packet belongs; said splitter determining, according to the correlations between the identification information and threads of the data stream in the flow table corresponding to the transport layer communications protocol to which said packet belongs, the thread corresponding to the data stream to which said packet belongs; said splitter sending said packet to the buffer queue of the thread corresponding to said data stream, to enable the thread corresponding to the data stream to obtain the packet from said buffer queue.

Description

technical field [0001] The invention relates to the technical field of computers, in particular to a data splitting method and a splitter. Background technique [0002] Today, processors have entered the multi-core era. Under the multi-core architecture, multi-threads are often deployed in multiple cores to achieve parallel processing. Then, it is necessary to send data packets to different threads by shunting to make multiple cores Implement parallel processing. [0003] In the existing data distribution method, the distribution granularity is process, that is, the data packet is sent to the buffer queue of the process corresponding to the data packet. Threads in this process obtain data packets from the buffer queue of this process, and the connection information of this data packet may be shared by other threads. If it is shared, when the thread accesses the connection information corresponding to the data packet, the mutual exclusion and synchronization mechanism betwe...

AI Technical Summary

Problems solved by technology

[0004] Therefore, the existing data shunting method has the following disadvantages: multiple threads share the buffer queue of the same process, which will cause a large amount of mutual exclusion and synchronization overhead between threads and cache failure between cores, resulting in insufficient processing capacity of multi-core processors. to play

Images