| 摘要 |
The present invention provides a technique for, in the case in which a failure has occurred in a shared memory, controlling a period of a pseudo through operation to reduce a period in which performance of a disk array device falls. Control information is divided into management information, which is required to be duplexed, and directory information, which is only required to simplexed, and the management information and the directory information are stored in separate shared memories. In the case in which a failure has occurred in the shared memory of an expanded memory unit (Option) storing the directory information, the directory information is reestablished in the shared memory of a basic memory unit (Basic). The pseudo through operation is cancelled at the point when the directory information is reestablished. After a package of the expanded memory unit is replaced with a normal product, the directory information is reestablished again. Management information managed in the other cluster is copied to the shared memory of the basis memory unit to complete maintenance and recovery work. |
