System and method of disaster recovery

• US 7,860,824 B2
• Filed: 05/21/2007
• Issued: 12/28/2010
• Est. Priority Date: 03/15/2007
• Status: Expired due to Fees

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Watanabe, Satoru, Mizuno, Kazuhiko, Fujiwara, Shinji, Suzuki, Yoshio, Kawamura, Nobuo
• Primary Examiner(s): Kim; Charles
• Assistant Examiner(s): Le; Jessica N
• Application Number:
  US11/802,186
• Publication Number:
  US 20080229140A1
• Time in Patent Office: 1,317 Days
• Field of Search: None
• US Class Current: 707/609
• CPC Class Codes: G06F 11/2074 Asynchronous techniques
• CPC Class Codes: G06F 11/2082 Data synchronisation
• CPC Class Codes: G06F 16/2358 Change logging, detection, and notification replication G06F16/27
• CPC Class Codes: G06F 2201/855 Details of asynchronous mirroring using a journal to transfer not-yet-mirrored changes

Abstract

In a DR system, from the viewpoint of device cost, when search is not carried out, a physical application where log recovery is available by inexpensive DB appliance server is adopted. Further, a local mirror operation at a secondary site is not carried out. Furthermore, from the viewpoint of operation, by a log apply function unit, the tendencies of a log application and operations are monitored, and a search process is accepted according to the progress conditions of the log application. When the log application does not catch up sufficiently, the search is not accepted. Moreover, when a consistency guarantee of a secondary DB is made, not transactions in process at the moment of search instruction are undone (rolled back), but only transactions in process at the moment of a search instruction are redone (rolled forward).

Claims

1. A log-based system for disaster recovery, the log-based system comprising:

a primary database server maintaining a database in a primary storage, generating, by a processor, log records describing modifications of the database, storing the log records in a log file formed in the primary storage, and transferring the log records to a secondary storage in which a remote copy of the log file is stored; and

a secondary database server storing a remote copy of the database in the secondary storage and maintaining the remote copy of the database according to the log records stored in the remote copy of the log file, the secondary database server including a log apply component and a search component wherein

the log apply component includes a log application unit applying the modifications described in the log records to the remote copy of the database stored in the secondary storage by sequentially reading out the log records from the remote copy of the log file,

the search component receives a request for a search of the remote copy of the database stored in the secondary storage and accessing the remote copy of the database to perform the search upon suspension of the log application unit, and

the log apply component includes a judge unit monitoring a progress condition of application of the modifications to the remote copy of the database by the log application unit based on the log record being generated by the primary database server and suspending operation of the log application unit upon recognizing that the progress condition is acceptable to thereby permit the search component to access the remote copy of the database to perform the search,

wherein the search component issues the request for the search to the judge unit

wherein the judge unit monitors the progress condition upon receiving the request for the search from the search component,

wherein the judge unit, upon recognizing that the progress condition is acceptable, ensures that the remote copy of the database is in a state in which consistency will be attained, calculates a suspend time period during which the log application unit is permitted to be suspended, and suspends the log application unit for the suspend time period to thereby permit the search component to access the remote copy of the database to perform the search,

wherein the judge unit restarts the log application unit upon completion of the search by the search component or stops the search and restarts the log application unit upon expiration of the suspend time period,

wherein the judge unit monitors the progress condition of the log application unit by calculating an application delay ratio according to a most recent modification applied to the remote copy of the database by the log application unit and a most recent log record stored in the log file, and

wherein the judge unit recognizes that the progress condition of the log application unit is acceptable when the application delay ratio is less than a predetermined threshold.

2. The log-based system for disaster recovery according to claim 1,

wherein the judge unit ensures that the remote copy of the database is in a state in which consistency will be attained by directing the log apply component to complete application of each uncommitted modification currently being applied to the remote copy of the database stored in the secondary storage at a time corresponding to when the judge unit recognizes that the progress condition is acceptable.

3. The log-based system for disaster recovery according to claim 1,

wherein the search component comprises a control unit and a search unit, and

wherein the control unit receives the request for the search of the remote copy of the database stored in the secondary storage, issues the request for the search to the judge unit, and directs the search unit to access the remote copy of the database to perform the search upon the log apply component having ensured that the remote copy of the database is in a state in which consistency will be attained.

4. The log-based system for disaster recovery according to claim 1,

wherein the judge unit calculates the suspend time period based on a most recent modification applied to the remote copy of the database by the log application unit, a most recent log record stored in the log file a predetermined acceptable amount of log application delay, that ensures the remote copy of the database is in a condition in which consistency will be attained, and a log record generating speed of the primary database server.

5. The log-based system for disaster recovery according to claim 3,

wherein the judge unit stops the search upon expiration of the suspend time period by transmitting a stop instruction to the control unit, and

wherein the control unit, upon receiving the stop instruction, directs the search unit to cancel performance of the search if the search unit has not completed the search and sends a search cancellation notification to the judge unit upon performance of the search by the search unit being canceled.

6. The log-based system for disaster recovery according to claim 1,

wherein the search component issues the request for the search to the judge unit,

wherein the judge unit, upon receiving the request for the search from the search component, monitors the progress condition to determine whether the progress condition is acceptable,

wherein the judge unit, upon determining that the progress condition is not acceptable, calculates a search availability time at which the progress condition is expected to be acceptable, sends a rejection notification including the search availability time to the search component, and continues monitoring the progress condition to determine whether the progress condition is acceptable,

wherein the judge unit, upon recognizing that the progress condition is acceptable, ensures that the remote copy of the database is in a state in which consistency will be attained, calculates a suspend time period during which the log application unit is permitted to be suspended, and suspends the log application unit for the suspend time period to thereby permit the search component to access the remote copy of the database to perform the search, and

wherein the judge unit restarts the log application unit upon expiration of the suspend time period.

7. The log-based system for disaster recovery according to claim 6,

wherein the search component comprises a control unit and a search unit,

wherein the control unit receives the request for the search of the remote copy of the database at the secondary database server, issues the request for the search to the judge unit, and receives the rejection notice from the judge unit upon the judge unit determining that the progress condition is not acceptable,

wherein the control unit, upon receiving the rejection notice, adds the request for the search to a search queue and issues the request for the search to the judge unit again upon reaching the search availability time, and

wherein the control unit directs the search unit to access the remote copy of the database to perform the search upon the judge unit having suspended the log application unit.

8. The log-based system for disaster recovery according to claim 1,

wherein the judge unit receives checkpoint logs and, upon receiving a checkpoint log, ensures that the remote copy of the database is in a state in which consistency will be attained, and, upon ensuring that the remote copy of the database is in a state in which consistency will be attained, sends a search availability notification to the search component, and

wherein the search component accepts the request for the search upon receiving the search availability notification and, upon accepting the request for the search, issues the request for the search to the judge unit.

9. The log-based system for disaster recovery according to claim 8,

wherein the judge unit ensures that the remote copy of the database is in a state in which consistency will be attained by directing the log apply component to complete application of each uncommitted modification currently being applied to the remote copy of the database at the secondary database server at a time corresponding to when the judge unit receives the checkpoint log.

10. The log-based system for disaster recovery according to claim 8,

wherein the judge unit monitors the progress condition upon receiving the request for the search from the search component, and

wherein the judge unit, upon recognizing that the progress condition is acceptable, calculates a suspend time period during which the log application unit is permitted to be suspended, and suspends the log application unit for the suspend time period to thereby permit the search component to access the remote copy of the database to perform the search.

11. The log-based system for disaster recovery according to claim 8,

wherein the search component comprises a control unit and a search unit,

wherein the control unit directs the search unit to access the remote copy of the database to perform the search upon the judge unit having suspended the log application unit for the suspend time period, and

wherein the judge unit restarts the log application unit upon completion of the search by the search component or stops the search and restarts the log application unit upon expiration of the suspend time period.

12. A computer-implemented method of log-based disaster recovery for a system in which a database is maintained in a primary storage by a first database server and a remote copy of the database is maintained in a secondary storage by a secondary database server, the method comprising:

generating log records describing modifications of the database;

storing the log records in a log file formed in the primary storage;

transferring the log records to a secondary storage in which a remote copy of the log file is stored;

applying the modifications described in the log records to the remote copy of the database stored in the secondary storage by sequentially reading out the log records from the remote copy of the log file;

receiving a request for a search of the remote copy of the database;

judging a progress condition of application of the modifications to the remote copy of the database based on the generated log records stored in the log file formed in the primary storage upon the receiving the request for the search to determine whether the progress condition is acceptable for performing the search;

suspending application of the modifications to the remote copy of the database upon recognizing that the progress condition is acceptable for performing the search; and

accessing the remote copy of the database to perform the search suspension of the application of the modifications to the remote copy of the database,

wherein suspending application of the modifications to the remote copy of the database comprises ensuring that the remote copy of the database is in a state in which consistency will be attained, calculating a suspend time period during which application of the modifications to the remote copy of the database is permitted to be suspended, and suspending application of the modifications to the remote copy of the database for the suspend time period, and

further comprising restarting application of the modifications to the remote copy of the database upon completion of the search or stopping performance of the search and restarting application of the modifications to the remote copy of the database upon expiration of the suspend time period;

wherein judging the progress condition of application of the modifications to the remote copy of the database comprises calculating an application delay ratio according to a most recent modification applied to the remote copy of the database and a most recent log record stored in the log file and recognizing that the progress condition is acceptable when the application delay ratio is less than a predetermined threshold.

13. The method according to claim 12,

wherein ensuring that the remote copy of the database is in a state in which consistency will be attained comprises completing application of each uncommitted modification currently being applied to the remote copy of the database at a time corresponding to when the progress condition is recognized as being acceptable.

14. The method according to claim 12,

wherein calculation of the suspend time period is based on a most recent modification applied to the remote copy of the database, a most recent log record stored in the log file, a log file output state of the primary database server, a predetermined acceptable amount of log application delay, and a log record generating speed.

15. The method according to claim 14,

further comprising cancelling performance of the search upon expiration of the suspend time period if performance of the search is not completed.

16. The method according to claim 12,

wherein suspending application of the modifications to the remote copy of the database comprises ensuring that the remote copy of the database is in a state in which consistency will be attained, calculating a suspend time period during which application of the modifications to the remote copy of the database is permitted to be suspended, and suspending application of the modifications to the remote copy of the database for the suspend time period,

further comprising, upon determining that the progress condition of application of the modifications to the remote copy of the database is not acceptable for performing the search, calculating a search availability time at which the progress condition is expected to be acceptable, and continuing judging the progress condition of application of the modifications to the remote copy of the database to determine whether the progress condition is acceptable for performing the search, and

further comprising restarting application of the modifications to the remote copy of the database upon expiration of the suspend time period.

17. The method according to claim 16,

further comprising, upon determining that the progress condition is not acceptable, adding the request for the search to a search queue and judging the progress condition of application of the modifications to the remote copy of the database to determine whether the progress condition is acceptable for performing the search upon reaching the search availability time.

18. The method according to claim 12,

further comprising receiving checkpoint logs and, upon receiving a checkpoint log, ensuring that the remote copy of the database is in a state in which consistency will be attained, and

further comprising, upon ensuring that the remote copy of the database is in a state in which consistency will be attained, calculating a suspend time period during which application of the modifications to the remote copy of the database is permitted to be suspended, and suspending application of the modifications to the remote copy of the database for the suspend time period.