Grandfather, father, son, mission critical, geographic location, restore point, checkpoint, forward recovery, backward recovery, recovery manager
Data Mill Backup tutorial
A backup is a copy of a file or a copy or an entire system in order that there is another copy incase the original is destroyed or corrupted. A DBMS has backup and recovery procedures to safeguard against data loss. These backup copies are normally made at a scheduled time each day. When a DBMS makes a backup copy of the database it is would also make a journal with a listing of all actions completed by the DBMS. The journal can then be checked against the data to double check data entries.
The storage location of backups is important. For instance if the backup is stored geographically close then it may also be destroyed in the case of a fire. If the data is mission critical it would normally be stored in a different geographic location. However there are different ways to backup and different goals.
This refers to a backup procedure where at least 3 generations of backups are stored. The most recent backup is called the son. The oldest is the grandafther. This means that a system can be restored to any of those three restore points. This is most commonly used with tape backups during batch processing but is still commonly used with any other system as well. It is just a smart thing to do. Apparently large organisations will keep more than three generations with banks keeping up to 20.
partial backups occur when only parts of the master file are modified to reflect changes to the system. The master file will be backed up at a regular interval such as daily or weekly. In between there maybe partial backups.
A checkpoint is where all processing is finalised and the computer checks that all information in the DBMS is synchronised. Checkpoints make it easier to know reliable restore points when checking journals and database logs. Checkpoints should be done several times an hour. If a system failure occurs processing can be resumed from the last checkpoint.
This is the program that restores the database to the nearest checkpoint and allows a system to resume operations.
This is used to undo unwanted changes. This involves the recovery manager restoring the system to an earlier checkpoint.
This occurs after there has already been problems and a system has been restored to an earlier checkpoint. The data is then re-entered to bring it back to where it was. The data may be re-entered from backups or manually using logs and journals.
Jounals contain all the essential changes to the database. There are two kinds: jounals and logs. Firstly, the logs contain all the essential data for a transaction in a transaction processing system where money is involved such as amount, date, type of transactions etc. This can normally be verified by by the till roll, copies of receipts given to customers. The second type is the journal which actually documents changes to the system and the database itself.