To shorten a performance degradation period of a disk array system when a shared memory fails, by restricting a pseudo through operation period.
Control information is divided into management information D1 that must be duplex, and directory information D2A that can be simplex, and both are stored in separate shared memories 61A and 61B. When the shared memory 61B of an expansion memory part (Option) storing the directory information D2A fails (S11), the directory information D2A is reconstructed in the shared memory 61A of a basic memory part (Basic). When the directory information D2A is reconstructed, a pseudo through operation is canceled. After the package of the expansion memory part is replaced with a normal one, the directory information D2A is reconstructed again (S22). The management information managed in the other cluster is copied into the shared memory 61A of the basic memory part (S26) to complete the maintenance and recovery operation.
OKUMOTO KATSUHIRO
HONMA HISAO
Miyagawa Nagao
Takeshi Nakamura
Next Patent: PATTERN DETECTING DEVICE AND PATTERN DETECTING METHOD