摘要 |
PROBLEM TO BE SOLVED: To provide a distributed shared memory system for efficiently processing large-scaled data mining such as a TB(tera byte) order. SOLUTION: When an item appearance frequency counting part 18 of a process 11 counts appearance frequency 15 of an item existing in a shared memory space 12 based on transaction data 17, a counting log transferring part 21 receives a counting log 20 recorded in a counting log recording part 19, and transfers it to another node 100. On the other hands, a counting log receiving part 22 transfers the counting log 20 transferred from the node 100 to a counting log reflecting part 23. Then, the counting log reflecting part 23 to which the counting log 20 is transferred reflects the content on the appearance frequency 15 of the item existing in the shared memory space 12. |