摘要 |
PROBLEM TO BE SOLVED: To improve the using efficiency of a resource for saving of a memory recovery processing necessitated at the failure occurrence of an application by automatically recovering memory until a memory updated state when a transaction issues a commit. SOLUTION: A memory managing part 2 updates a memory update history file 3 and a memory part 4 in updating a memory, and refers to only the memory part 4 in referring to a memory based on memory update/reference information, and returns the result to a network application 1. Then, a memory update history file 3 is actually updated when a commit is issued in the network application 1. When system down or transaction back-out is caused, the network application 1 reports it to a memory managing part 2 at the transaction re-input. The memory managing part 2 which receives this report develops the memory update information until just before the commit from the memory update history file 3 to the memory part 4. |