Detailed analysis of different recovery features in Oracle database

Since Oracle database backups are available in three ways, each with different recovery characteristics, you should integrate database and file system backups, integrate logical and physical backups.

First, backup strategy

1. At the operating system level, with a large-capacity disk array, each database file is automatically distributed to each physical disk through disk image technology. In this way, when a disk is physically damaged, the operating system automatically raises the image disk to replace the failed disk, ensuring that the database is running properly.

Because we are using a dual server, keep a backup database on another server. The backup database has the same parameter state as the database in use, so that in the case of a database failure, only minimal recovery is required to minimize recovery time.

Maintain multiple backup of control files on multiple different physical disks. Control files are used to boot Oracle during database recovery, so maintaining multiple backup of control files ensures that the available control files can be used for database recovery after a disk failure occurs.

Make the database run in Archivelog (archived) mode, and the archive log is stored on a logical disk in another image.

Perform an online backup operation every night, backing up all data files, all archived log files, and a control file.

An export operation is performed once a week.

Ii. Recovery Strategy

1. Instance failure

Recovery from an instance failure is automatic. Instance failures are typically caused by a server failure, restarting the server after the database instance fails, starting the database, Oracle checking data files and online log files, and synchronizing all files to the same point in time.

2. Disk failure

If the loss is a control file, just close the database and copy it from the place where the control file is kept.

If you lose a data file, you can recover it using an online backup from the previous night, as follows:

(1) From the backup to the original location of the missing files.

(2) Loading the database

(3) Restore the database

(4) Open the database

3. Delete or modify an object incorrectly

In this case, it is generally desirable to return back to the point in time before the error occurred. This is called Point-in-time recovery. The steps to complete the recovery are as follows:

(1) from the current database output (export), input (import) to the standby database, so that the standby database and the current database to maintain the same.

(2) Roll forward the standby database to the point of time before the error occurred. ---www.bianceng.cn

(3) Output the logical object affected by the error from the standby database.

(4) Enter those affected objects into the current database using the output file produced in the previous step.

Third, User role management

1. Classify all clients by job nature, and grant different user roles respectively.

2. To different user roles, according to their use of data sources, respectively, grant different access to the database objects.

Based on the above security policy, can prevent illegal users access to the database, limit the permissions of legitimate users to manipulate the database, you can make the database in the event of failure, will not lose any data, and can make the database quickly restored to the point before the failure of the time, to maximize the security of data and the continuous operation of the entire system.