Before recovery is performed, you need to restore the database. To restore means to bring back an original copy of a file from backup by issuing the operating system
command. To recover a database means to make all the restored files current to the same point in time by issuing the
command. To do this, you apply all or part of the redo information generated by the LGWR (Log Writer
process since the last backup. LGWR records this information in the online redo logs that are archived when the database is operating in ARCHIVELOG mode. The damaged datafiles and control files can be recovered to the last recorded
before the media failure.
The process of recovering datafiles and control files by applying the redo logs is also known as "rolling forward." A minimum of two online redo logs is required by LGWR. Each log switch causes a new log sequence number to be used and assigned to the current log file.
For a successful recovery, certain data structures must be in place. The data structure must include
- redo records,
- rollback segments,
- redo logs, and
- a control file.
These data structures apply to both NOARCHIVELOG and ARCHIVELOG database recovery.
The next lesson describes the data structures required to restore a NOARCHIVELOG database.