Nonstop Database, HiRDB Version 9 Disaster Recovery System Configuration and Operation Guide
If you perform any of the operations described in Table 4-4 or Table 4-5 when using the hybrid method, the data that was updated during that operation cannot be recovered from the system log file. Therefore, the databases on the main site and the remote site need to be re-synchronized. HiRDB performs the actual synchronization process, but the resulting overhead causes the processing time for the operations described in Table 4-4 or Table 4-5 to be longer than normal. Furthermore, if the KFPS04680-E error message is output when performing these operations, data recovery cannot be guaranteed for the RDAREAs updated by these operations after switching sites. Table 4-4 shows the commands that require the databases to be re-synchronized and Table 4-5 shows the operations that require the databases to be re-synchronized.
Table 4-4 Commands that require the databases to be re-synchronized
| Command name | Option | Command description | Overhead | RDAREA for which data recovery is not guaranteed if the KFPS04680-E message is output |
|---|---|---|---|---|
| pdhold | -s | Synchronization hold | 2 (seconds for each RDAREA specified in the -r option) | RDAREA specified in the -r option |
| -c | Command hold and closure | |||
| pdclose | -- | RDAREA closure | ||
| pdrels | -- | RDAREA hold release | ||
| pdorbegin | -r | Committing an online-reorganized database | 2 (seconds for each RDAREA specified in the -r option) | RDAREA specified in the -r option |
| -s | 2 (seconds for each RDAREA specified in the -s option) | RDAREA of the server specified in the -s option | ||
| -t | 2 (seconds for each RDAREA storing the table specified in the -t option) | RDAREA storing the table specified in the -t option | ||
| pdorend | When the -s option is omitted | Reflection processing of online reorganization | 2 (seconds for each RDAREA that is in the online reorganization hold state) | RDAREA that is in the online reorganization hold state |
| -s | 2 (seconds for each RDAREA that is in the online reorganization hold state out of the RDAREAs of the server specified in the -s option) | RDAREA that is in the online reorganization hold state out of the RDAREAs of the server specified in the -s option |
Table 4-5 Operations that require the databases to be re-synchronized
| Operation | Overhead | RDAREA for which data recovery is not guaranteed if the KFPS04680-E error message is output | |
|---|---|---|---|
| Commit or rollback | Updating using HiRDB Text Search Plug-in | 2 (seconds for each updated RDAREA) | User LOB RDAREAs targeted for updating |
| Updating of BLOB data | |||
| Updating using pre-update log acquisition mode or no-log mode | User RDAREAs and user LOB RDAREAs targeted for updating | ||
| Updating of shared RDAREAs | Shared RDAREAs targeted for updating | ||
| Updating of RDAREAs whose open attribute is SCHEDULE | User RDAREAs and user LOB RDAREAs targeted for updating | ||
| DISCONNECT | Updating of shared RDAREAs | Shared RDAREAs targeted for updating | |
| Execution of definition SQL on shared table# | 2 seconds | Shared RDAREA on which definition SQL is executed | |
| Execution of LOCK TABLE with lock mode specification on shared table# | Shared RDAREA on which LOCK TABLE is executed | ||
| Automatic extension of RDAREAs | 2 (seconds for each operation) | RDAREA for which automatic extension is executed | |
#: Applicable to a HiRDB/Parallel Server, but not to a HiRDB/Single Server.
All Rights Reserved. Copyright (C) 2011, Hitachi, Ltd.