LOGDB and TRANSDB Importance
The transaction information in the LOGDB system databank is vitally important because it is used in the event of a disk crash to fully recover a destroyed databank.
The LOGDB information will contain data on all the changes made to the databank from the time the backup copy of the databank was taken until the time of the disk crash. This information is used if a backup copy of a databank file is to be restored.
Note: Data changes that are not logged cannot be restored by this process, therefore it is important to consider the issue of transaction logging carefully.
If a databank becomes unavailable (because the Mimer SQL system is stopped deliberately or by a system failure) during the commitment of a transaction, information is retained in the TRANSDB system databank and used to complete the transaction when the databank becomes available again.
This is only true for databanks with the TRANS or LOG option. Once information has been successfully written to both LOGDB and the databank file, it is removed from TRANSDB.
It is recommended practice to back up all the databanks of the database at the same time, and to ensure that LOGDB is always backed up whenever other databanks are backed up, because the LOGDB information provides the transaction data which links the previous backup copy of a databank with the databank as it exists at the current point in time.
Thus, when restoring a databank it should be brought to a state consistent with the latest backup. This is done either by using the latest backup copy of the databank or by using backed up LOGDB information with an older backup copy of the databank.
The current LOGDB system databank is then used to restore the final changes made between the time of the latest backup and the time the databank was lost.
The graphic above describes a scenario which ends up in a system crash.
To recover from this situation the common operation is to start from the most recent backup (T2) and then use the current LOGDB to recover data up to the state at T3.
When the system is restarted, the current TRANSDB is used to automatically recover up to the moment of the crash.
If the most recent backup cannot be used, an older backup has to be brought in (T1). This backup is restored up to the consistent state at T2 by using the LOGDB stored in the backup at T2.
Then the current LOGDB and TRANSDB are used to restore the transactions committed after the backup at T2.
Note: Wherever possible, LOGDB should be stored on a different disk unit, with a separate disk controller, from the other databanks in order to minimize the risk that a disk crash or damaged disk controller destroys both the log and the other databanks.
LOGDB and TRANSDB should always be located on different physical disks which are ideally served by separate disk controllers and no other databank files should be located on either disk, since data may be lost if both TRANSDB and LOGDB are destroyed.
Refer to Organizing Databank Files, for more details on data security and databank files.
Upright Database Technology AB
Voice: +46 18 780 92 00
Fax: +46 18 780 92 40