Oracle Backup and Recovery

Backup and recovery is one of the most important aspects of database administration. If a database crashed and there was no way to recover it, the devastating results to a business could include lost data, lost revenue and customer dissatisfaction. Whether companies operate a single database or multiple databases storing hundreds of gigabytes or even terabytes of data, they share one common factor - the need to back up important data and protect themselves from disaster by developing a backup and recovery plan.

A backup is a representative copy of data. This copy can include important parts of a database such as the control file, redo logs, and datafiles. A backup protects data from application error and acts as a safeguard against unexpected data loss, by providing a way to restore original data. Backups are divided into physical backups and logical backups. Physical backups are copies of physical database files. The phrase "backup and recovery" usually refers to the transfer of copied files from one location to another, along with the various operations performed on these files.

In contrast, logical backups contain data that is exported using SQL commands and stored in a binary file. Oracle records both committed and uncommitted changes in redo log buffers. Logical backups are used to supplement physical backups. Restoring a physical backup means reconstructing it and making it available to the Oracle server. To recover a restored backup, data is updated using redo records from the transaction log. The transaction log records changes made to the database after the backup was taken.

Oracle performs crash recovery and instance recovery automatically after an instance failure. In the case of media failure, a database administrator (DBA) must initiate a recovery operation. Recovering a backup involves two distinct operations: rolling the backup forward to a more recent time by applying redo data, and rolling back all changes made in uncommitted transactions to their original state. In general, recovery refers to the various operations involved in restoring, rolling forward, and rolling back a backup. Backup and recovery refers to the various strategies and operations involved in protecting the database against data loss and reconstructing the database should a loss occur.

A backup is a snapshot of a datafile, tablespace, or database at a certain time. If periodic backups of the database have been made and data is lost, users can apply the stored redo information to their latest backup to make the database current again. Oracle enables users to restore an older backup and apply only some redo data, thereby recovering the database to an earlier point in time. This type of recovery is called incomplete media recovery. If the backup was consistent, then users are not required to apply any redo data, at all.

A simple example of media recovery illustrates the concept. Suppose a user makes a backup of the database at noon. Starting at noon, one change to the database is made every minute. At 1 p.m. one of the disk drives fails, causing the loss of all data on that disk. Fortunately, Oracle records all changes in the redo log. The user can then restore the noon backup onto a functional disk drive and use redo data to recover the database to 1 p.m., reconstructing the lost changes.

Although backup and recovery operations can be intricate and vary from one business to another, the basic principles follow these four simple steps:

  1. Multiplex the online redo logs
  2. Run the database in ARCHIVELOG mode and archive redo logs to multiple locations
  3. Maintain multiple concurrent backups of the control file
  4. Take frequent backups of physical datafiles and store them in a safe place, making multiple copies if possible

As long as users have backups of the database and archive redo logs in safe storage, the original database can be recreated.

Before users begin to think seriously about backup and recovery strategy, the physical data structures relevant for backup and recovery operations must be identified. This section discusses the following physical data structures:

Every Oracle database has one or more physical datafiles that belong to logical structures called tablespaces. The datafile is divided into smaller units called data blocks. The data of logical database structures, such as tables and indexes, is physically located in the blocks of the datafiles allocated for a database. Datafiles hold the following characteristics:

The first block of every datafile is the header. The header includes important information such as file size, block size, tablespace, and creation timestamp. Whenever the database is opened, Oracle checks to see that the datafile header information matches the information stored in the control file. If it does not, then recovery is necessary. Oracle reads the data in a datafile during normal operation and stores it in the buffer cache. For example, assume that a user wants to access some data in a table. If the requested information is not already in the buffer cache, Oracle reads it from the appropriate datafiles and stores it in memory.

Every Oracle database has a control file containing the operating system filenames of all other files that constitute the database. This important file also contains consistency information that is used during recovery, such as the:

Users can multiplex the control file, allowing Oracle to write multiple copies of the control file to protect it against disaster. If the operating system supports disk mirroring, the control file can also be mirrored, allowing the O/S to write a copy of the control file to multiple disks. Every time a user mounts an Oracle database, its control file is used to identify the datafiles and online redo log files that must be opened for database operation. If the physical makeup of the database changes, such as a new datafile or redo log file is created, Oracle then modifies the database's control file to reflect the change. The control file should be backed up whenever the structure of the database changes. Structural changes can include adding, dropping, or altering datafiles or tablespaces and adding or dropping online redo logs.

Redo logs are absolutely crucial for recovery. For example, imagine that a power outage prevents Oracle from permanently writing modified data to the datafiles. In this situation, an old version of the data in the datafiles can be combined with the recent changes recorded in the online redo log to reconstruct what was lost. Every Oracle database contains a set of two or more online redo log files. Oracle assigns every redo log file a log sequence number to uniquely identify it. The set of redo log files for a database is collectively known as the database's redo log.

Oracle uses the redo log to record all changes made to the database. Oracle records every change in a redo record, an entry in the redo buffer describing what has changed. For example, assume a user updates a column value in a payroll table from 5 to 7. Oracle records the old value in undo and the new value in a redo record. Since the redo log stores every change to the database, the redo record for this transaction actually contains three parts:

If the user then commits the update to the payroll table - to make permanent the changes executed by SQL statements - Oracle generates another redo record. In this way, the system maintains a careful watch over everything that occurs in the database.

Circular Use of Redo Log Files
Log Writer (LGWR) writes redo log entries to disk. Redo log data is generated in the redo log buffer of the system global area. As transactions commit and the log buffer fills, LGWR writes redo log entries into an online redo log file. LGWR writes to online redo log files in a circular fashion: when it fills the current online redo log file, called the active file, LGWR writes to the next available inactive redo log file. LGWR cycles through the online redo log files in the database, writing over old redo data. Filled redo log files are available for reuse depending on whether archiving is enabled:

Archived log files are redo logs that Oracle has filled with redo entries, rendered inactive, and copied to one or more log archive destinations. Oracle can be run in either of two modes:

Running the database in ARCHIVELOG mode has the following benefits:

Running the database in NOARCHIVELOG mode has the following consequences:

Automatic Managed Undo
Every Oracle database must have a method of maintaining information that is used to roll back, or undo, changes to the database. Such information consists of records of the actions of transactions, primarily before they are committed. Oracle refers to these records collectively as undo. Historically, Oracle has used rollback segments to store undo. Space management for these rollback segments has proven to be quite complex. Oracle now offers another method of storing undo that eliminates the complexities of managing rollback segment space, and allows DBAs to exert control over how long undo is retained before being overwritten. This method uses an undo tablespace. Undo records are used to:

When a rollback statement is issued, undo records are used to undo changes that were made to the database by the uncommitted transaction. During database recovery, undo records are used to undo any uncommitted changes applied from the redo log to the datafiles. Undo records provide read consistency by maintaining the before image of the data for users who are accessing the data at the same time that another user is changing it.

Understanding Basic Backup
A backup strategy provides a safeguard against data loss. Answering the following questions can help database administrators develop a strong backup strategy:

Data loss can occur for various reasons. Here are some of the most common types of failures that can lead to data loss.

A statement failure is a logical failure in the handling statement in an Oracle program. For example, a user issues a statement that is not a valid SQL construction. When statement failure occurs, Oracle automatically undoes any effects of the statement and returns control to the user.

A process failure is a failure in a user process accessing Oracle, i.e., an abnormal disconnection or process termination. The failed user process cannot continue work, although Oracle and other user processes can. If the user process fails while modifying the database, Oracle background processes undo the effects of uncommitted transactions.

An instance failure is a problem that prevents an Oracle instance, i.e., the SGA and background processes, from continuing to function. Instance failure can result from a hardware problem such as a power outage, or a software problem such as an operating system crash. When an instance fails, Oracle does not write the data in the buffers of the SGA to the datafiles.

A user or application error is a user mistake that results in the loss of data. For example, a user can accidentally delete data from a payroll table. Such user errors can require the database or object to be recovered to a point in time before the error occurred. To allow recovery from user error and accommodate other unique recovery requirements, Oracle provides Flashback Technology. For more information on Flashback Technology.

A media failure is 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. Disk failure can affect a variety of files, including datafiles, redo log files, and control files. Because the database instance cannot continue to function properly, it cannot write the data in the database buffers of the SGA to the datafiles.

A database contains a wide variety of types of data. When developing backup strategy, DBAs must decide what information they want to copy. The basic backup types include:

  • Online Database Backup
  • Offline Database Backup
  • Whole Database
  • Tablespace
  • Datafile
  • Control File
  • Archived Redo Log
  • Configuration Files

In deciding what to back up, the basic principle is to prioritize data depending on its importance and the degree to which it changes. Archive logs do not change, for example, but they are crucial for recovering the database, so multiple copies should be maintained, if possible. Expense account tables, however, are constantly updated by users. Therefore, this tablespace should be backed up frequently to prevent having to apply as much redo data during recovery.

Backups can be combined in a variety of ways. For example, a DBA can decide to take weekly whole database backups, to ensure a relatively current copy of original database information, but take daily backups of the most accessed tablespaces. The DBA can also multiplex the all important control file and archived redo log as an additional safeguard.

Online Database Backup
An online backup or also known as an open 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 other read-write datafile headers contain an SCN of 95 or 90. Oracle cannot open the database until all of these header SCNs are consistent, that is, until all changes recorded in the online redo logs have been saved to the datafiles on disk. If the database must be up and running 24 hours a day, 7 days a week, then you have no choice but to perform online backups of a whole database which is in ARCHIVELOG mode.

Offline Database Backup
In this backup, all datafiles and control files are consistent to the same point in time - consistent with respect to the same SCN, for example. The only tablespaces in a consistent backup that are allowed to have older SCNs are read-only and offline-normal tablespaces, which are consistent with the other datafiles in the backup. This type of backup allows the user to open the set of files created by the backup without applying redo logs, since the data is already consistent. The only way to perform this type of backup is to shut down the database cleanly and make the backup while the database is closed. A consistent whole database backup is the only valid backup option for databases running in NOARCHIVELOG mode.

Whole Database Backup
The most common type of backup, a whole database backup contains the control file along with all database files that belong to a database. If operating in ARCHIVELOG mode, the DBA also has the option of backing up different parts of the database over a period of time, thereby constructing a whole database backup piece by piece.

Tablespace Backups
A tablespace backup is a subset of the database. Tablespace backups are only valid if the database is operating in ARCHIVELOG mode. The only time a tablespace backup is valid for a database running in NOARCHIVELOG mode is when that tablespace is read-only or offline-normal.

Datafile Backups
A datafile backup is a backup of a single datafile. Datafile backups, which are not as common as tablespace backups and are only valid if the database is run in ARCHIVELOG mode. The only time a datafile backup is valid for a database running in NOARCHIVELOG mode is if that datafile is the only file in a tablespace. For example, the backup is a tablespace backup, but the tablespace only contains one file and is read-only or offline-normal.

Control File Backups
A control file backup is a backup of a database's control file. If a database is open, the user can create a valid backup by issuing the following SQL statement: ALTER DATABASE BACKUP CONTROLFILE to 'location'; or use Recovery Manager (RMAN).

Archived Redo Log Backups
Archived redo logs are the key to successful media recovery. Depending on the disk space available and the number of transactions executed on the database, you want to keep as many days of archive logs on disk and you want to back them up regularly to ensure a more complete recovery.

Configuration Files
Configuration files may consist of spfile or init.ora, password file, tnsnames.ora, and sqlnet.ora. Since these files do not change often, then they require a less frequent backup schedule. If you lost a configuration file it can be easily recreated manually. When restore time is a premium, it will be faster to restore a backup of the configuration file then manually creating a file with a specific format.

Oracle provides users a choice of several basic methods for making backups. The methods include:

  • Recovery Manager (RMAN) - A component that establishes a connection with a server process and automates the movement of data for backup and recovery operations.
  • Oracle Enterprise Manager - A GUI interface that invokes Recovery Manager.
  • Oracle Data Pump - The utility makes logical backups by writing data from an Oracle database to operating system files in a proprietary format. This data can later be imported into a database.
  • User Managed - The database is backed up manually by executing commands specific to the user's operating system.

Making Recovery Manager Backups
Recovery Manager (RMAN) is a powerful and versatile program that allows users to make an RMAN backup or image copy of their data. When the user specifies files or archived logs using the RMAN BACKUP command, By default RMAN creates a backup set as output. A backup set is a file or files in a proprietary-specific format that requires the use of the RMAN RESTORE command for recovery operations. In contrast, when the BACKUP AS COPY command is used to create an image copy of a file, it is in an instance-usable format - the user does not need to invoke Recovery Manager to restore or recover it.

When a RMAN command is issued, such as backup or restore, Recovery Manager establishes a connection to an Oracle server process. The server process then back up the specified datafile, control file, or archived log from the target database. The recovery catalog is a central repository containing a variety of information useful for backup and recovery. RMAN automatically establishes the names and locations of all the files needed to back up. Recovery Manager also supports incremental backups - backups of only those blocks that have changed since a previous backup. In traditional backup methods, all the datablocks ever used in a datafile must be backed up.

Automatic Disk-Based Backup and Recovery
The components that creates different backup and recovery-related files have no knowledge of each other or of the size of the file systems where they store their data. With Automatic Disk-Based Backup and Recovery, you can create a flash recovery area, which automates management of backup-related files. Choose a location on disk and an upper bound for storage space, and set a retention policy that governs how long backup files are needed for recovery, and the database manages the storage used for backups, archived redo logs, and other recovery-related files for your database within that space. Files no longer needed are eligible for deletion when RMAN needs to reclaim space for new files. If you do not use a flash recovery area, you must manually manage disk space for your backup-related files and balance the use of space among the different types of files. Oracle Corporation recommends that you enable a flash recovery area to simplify your backup management.

For more information on RMAN

Oracle Enterprise Manager
Although Recovery Manager is commonly used as a command-line utility, the Backup Wizards in Oracle Enterprise Manager is the GUI interface that enables backup and recovery via a point-and-click method. Oracle Enterprise Manager (EM) supports Backup and Recovery features commonly used by users.

  • Backup Configurations to customize and save commonly used configurations for repeated use
  • Backup and Recovery wizards to walk the user through the steps of creating a backup script and submitting it as a scheduled job
  • Backup Job Library to save commonly used Backup jobs that can be retrieved and applied to multiple targets
  • Backup Management to view and perform maintenace on RMAN backups.

For more information on Oracle Enterprise Manager

Using the Data Pump for Supplemental Backup Protection
Physical backups can be supplemented by using the Data Pump utility to make logical backups of data. Logical backups store information about the schema objects created for a database. Data Pump writes data from a database into Oracle files in a proprietary format, which can then be imported into a database using the Import utility.

For more information on Data Pump

User Managed Backups
Operating system commands can be used such as the UNIX dd or tar command to make backups. Backup operations can also be automated by writing scripts. The user can make a backup of the whole database at once or back up individual tablespaces, datafiles, control files, or archived logs. A whole database backup can be supplemented with backups of individual tablespaces, datafiles, control files, and archived logs. O/S commands can also be used to perform these backups if the database is down or if the database is placed into hot backup mode to take an online backup.

Basic recovery involves two parts: restoring a physical backup and then updating it with the changes made to the database since the last backup. The most important aspect of recovery is making sure all data files are consistent with respect to the same point in time. Oracle has integrity checks that prevent the user from opening the database until all data files are consistent with one another. When preparing a recovery strategy, it is critical to understand the answers to these questions:

  • How does recovery work?
  • What are the types of recovery?
  • Which recovery method should be used?

In every type of recovery, Oracle sequentially applies redo data to data blocks. Oracle uses information in the control file and datafile headers to ascertain whether recovery is necessary. Recovery has two parts: rolling forward and rolling back. When Oracle rolls forward, it applies redo records to the corresponding data blocks. Oracle systematically goes through the redo log to determine which changes it needs to apply to which blocks, and then changes the blocks. For example, if a user adds a row to a table, but the server crashes before it can save the change to disk, Oracle can use the redo record for this transaction to update the data block to reflect the new row.

Once Oracle has completed the rolling forward stage, the Oracle database can be opened. The rollback phase begins after the database is open. The rollback information is stored in transaction tables. Oracle searches through the table for uncommitted transactions, undoing any that it finds. For example, if the user never committed the SQL statement that added the row, then Oracle will discover this fact in a transaction table and undo the change.

There are three basic types of recovery: instance recovery, crash recovery, and media recovery. Oracle performs the first two types of recovery automatically at instance startup. Only media recovery requires the user to issue commands. An instance recovery, which is only possible in an Oracle Real Applications Cluster configuration, occurs in an open database when one instance discovers that another instance has crashed. A surviving instance automatically uses the redo log to recover the committed data in the database buffers that was lost when the instance failed.Oracle also undoes any transactions that were in progress on the failed instance when it crashed, then clears any locks held by the crashed instance after recovery is complete.

A crash recovery occurs when either a single-instance database crashes or all instances of a multi-instance database crash. In crash recovery, an instance must first open the database and then execute recovery operations. In general, the first instance to open the database after a crash or SHUTDOWN ABORT automatically performs crash recovery.

Unlike crash and instance recovery, a media recovery is executed on the user's command, usually in response to media failure. In media recovery, online or archived redo logs can be used to make a restored backup current or to update it to a specific point in time. Media recovery can restore the whole database, a tablespace or a datafile and recover them to a specified time. Whenever redo logs are used or a database is recovered to some non-current time, media recovery is being performed.

A restored backup can always be used to perform the recovery. The principal division in media recovery is between complete and incomplete recovery. Complete recovery involves using redo data combined with a backup of a database, tablespace, or datafile to update it to the most current point in time. It is called complete because Oracle applies all of the redo changes to the backup. Typically, media recovery is performed after a media failure damages datafiles or the control file.

For more information on Automatic Fault Recovery

Recovery Options
If the user does not completely recover the database to the most current time, Oracle must be instructed how far to recover. The user can perform:

  • Tablespace point-in-time recovery (TSPITR), which enables users to recover one or more tablespaces to a point-in-time that is different from the rest of the database.
  • Time-based recovery, also called point-in-time recovery (PITR), which recovers the data up to a specified point in time.
  • Cancel-based recovery, which recovers until the CANCEL command is issued.
  • Change-based recovery or log sequence recovery. If O/S commands are used, change-based recovery recovers up to a specified SCN in the redo record.
  • Flashback from human error

If Recovery Manager is used, log sequence recovery recovers up to a specified log sequence number. When performing an incomplete recovery, the user must reset the online redo logs when opening the database. The new version of the reset database is called a new incarnation. Opening the database with the RESETLOGS option tells Oracle to discard some redo. In Oracle Database 10g and following releases, the control file added new structures that provides the ability to recover through a RESETLOGS operation using backups from a previous incarnation.

Recovering From Human Errors
The Oracle Database 11g architecture leverages the unique technological advances in the area of database recovery due to human errors. Flashback Technology provides a set of new features to view and rewind data back and forth in time. The Flashback features offer the capability to query past versions of schema objects, query historical data, perform change analysis or perform self-service repair to recover from logical corruptions while the database is online. With the Oracle Database 11g Flashback Technology, you can indeed undo the past!

For more information on Flashback Technology

Users have a choice between two basic methods for recovering physical files. They can:

  • Use Recovery Manager to automate recovery.
  • Execute SQL commands.

Recovering with Recovery Manager
The basic RMAN commands are RESTORE and RECOVER. RMAN can be used to restore datafiles from backup sets or image copes, either to their current location or to a new location. If any archived redo logs are required to complete the recovery operation, RMAN automatically restores and applies them. In a recovery catalog, RMAN keeps a record containing all the essential information concerning every backup ever taken. If a recovery catalog is not used, RMAN uses the control file for necessary information. The RMAN RECOVER command can be used to perform complete media recovery and apply incremental backups, and to perform incomplete media recovery.

For more information on RMAN

Recovering with SQL*Plus
Administrators can use the SQL*Plus utility at the command line to restore and perform media recovery on your files. Users can recover a database, tablespace, or datafile. Before performing recovery, users need to:

  • Determine which files to recover. Often the table V$RECOVER_FILE can be used.
  • Restore backups of files permanently damaged by media failure. If the user does not have a backup, recovery can still be performed if the user has the necessary redo log files and the control file contains the name of the damaged file.
  • If a file cannot be restored to its original location, then the user must relocate the restored file and inform the control file of the new location.
  • Restore necessary archived redo log files.

Backup and recovery of your Oracle database is important to protecting data from corruptions, hardware failures, and data failures. While Oracle provides many features to protect your data, nothing can replace a backup.