Method or system for accessing and storing stream records of massive concurrent TCP streams

The invention discloses a method and a system for accessing and storing stream records of massive concurrent TCP streams. The method comprises the following steps: constructing a two-dimensional TCP stream record table, a stream record index table and TCP stream record buffer memory; selecting a message FIFO queue corresponding to a storage queue for storing the message according to the received tetrad Hash operation result of the message, and updating the stream record index table; accessing the message FIFO queue in a polling mode so as to read the message to be processed; accessing the stream record index table and the TCP stream record buffer memory according to the tetrad Hash operation result of the current message to be processed, and judging whether a TCP stream record related to the current message to be processed exists in the TCP stream record buffer memory, and directly accessing the TCP stream record buffer memory if yes; otherwise accessing the TCP stream record table according to the sequence i