초록 ▼

A database management system ensures consistency between primary and mirrored backup copies of a database, despite occurrence of a suspending condition interrupting the normal process of mirroring the primary database. One or more primary controllers are provided, each having a data storage unit wit

A database management system ensures consistency between primary and mirrored backup copies of a database, despite occurrence of a suspending condition interrupting the normal process of mirroring the primary database. One or more primary controllers are provided, each having a data storage unit with multiple primary data storage devices. Multiple secondary controllers each have multiple associated secondary data storage devices, each secondary controller being coupled to one primary controller. One or more primary databases reside on the primary devices, with a corresponding number of secondary databases residing on the secondary devices. Each secondary database mirrors a corresponding primary database. Either a host, attached to a primary controller, or one of the primary controllers itself, maintains a map cross-referencing each primary and secondary database with the primary and secondary devices containing portions thereof. If a predefined "suspending condition" affecting data mirroring occurs, the host or primary controller consults its map to identify all primary and secondary devices affected by the condition. Then, each primary controller stops all ongoing and future read/writes with each of its affected primary devices. Each primary controller also directs each secondary controller having an affected secondary device to stop mirroring the primary databases stored on that device. Then, the primary controller starts intermediate change recording and resumes read/writes with its primary devices. When the suspending condition ends, each primary controller applies the appropriate logged changes to its secondary database(s) and then reactivates each secondary database.

대표청구항 ▼

[What is claimed is:] [1.]at a first time, detecting unavailability of a first database for read/writes, said first database being a member of a first mirrored pair including the first database and a counterpart database;identifying all primary and secondary data storage devices containing any data

[What is claimed is:] [1.]at a first time, detecting unavailability of a first database for read/writes, said first database being a member of a first mirrored pair including the first database and a counterpart database;identifying all primary and secondary data storage devices containing any data of the first mirrored pair of databases, said identified primary and secondary data storage devices constituting affected primary and secondary devices;discontinuing all read/write operations by the affected primary devices ongoing at the first time, also preventing future read/write by the affected primary devices;instructing each secondary controller coupled to any affected secondary devices to terminate updating of the affected secondary devices;detecting termination of updating by the secondary controllers, and in response thereto, resuming read/writes with the affected primary devices;each primary controller coupled to affected primary devices receiving changes to the primary database of the first mirrored pair, and in response thereto, logging the received changes;detecting removal of the suspending condition, and in response thereto, resynchronizing the secondary database of the mirrored pair by:applying the logged changes to the secondary devices containing the secondary database of the mirrored pair; andinstructing the secondary controllers to resume updating of the secondary database of the mirrored pair to mirror its corresponding primary database.