Physical backup is a copy of a datafile, tablespace, or database made at a specific time. You can make physical backups using either the Oracle8i Recovery Manager utility or operating system utilities.

Logical backups contain data that you extract using the Oracle Export utility and store in a binary file. You can use logical backups to supplement physical backups.

A consistent backup of a database or part of a database is a backup in which all read-write datafiles and control files have been checkpointed with respect to the same SCN. In addition, all the online, read-write datafiles are not fuzzy, that is, do not contain changes beyond the SCN in the header.

An inconsistent backup is a backup in which all read-write datafiles and control files have not been checkpointed with respect to the same SCN. For example, one read-write datafile header may contain an SCN of 100 while others contain an SCN of 95.

The control file also contains crucial checkpoint information. The checkpoint SCN records the highest SCN of all changes to blocks such that all data blocks with changes below that SCN have been written to disk by DBWn. The control file also contains a record of the checkpoint SCN contained in the header of each datafile in the database. Whenever a discrepancy occurs between the SCN that is actually in a datafile header and the datafile header SCN listed in the control file, Oracle requires media recovery.

 Instance recovery is an automatic procedure that involves two distinct operations: rolling forward the backup to a more current time by applying online redo records and rolling back all changes made in uncommitted transactions to their original state.

 In contrast to instance recovery, media recovery requires you to issue recovery commands. If you use SQL*Plus, then you can issue the RECOVER or ALTER DATABASE RECOVER statements to apply the archived logs.

Every time a change is made, Oracle generates a record of both the changed and original value in the redo log buffer in memory. This record is called a redo record. Oracle records both committed and uncommitted changes in redo log buffers.

Each atomic change in a redo record is called a change vector. A redo record is composed of all the change vectors involved in a change. For example, if you update all the values in a multi-column row in a table, then Oracle generates a redo record containing change vectors for all the changed blocks corresponding to each updated value.

The system change number (SCN) is an ever-increasing internal timestamp that uniquely identifies a committed version of the database. Every time a user commits a transaction, Oracle records a new SCN.

Oracle uses SCNs in control files, datafile headers, and redo records. Every redo log file has both a log sequence number and low and high SCN. The low SCN records the lowest SCN recorded in the log file, while the high SCN records the highest SCN in the log file.

A problem that prevents an Oracle instance, that is, the SGA and background processes, from continuing to function.

A physical problem that arises when Oracle tries to write or read a file that is required to operate the database. A common example is a disk head crash that causes the loss of all data on a disk drive.