10.4. Summary

In this chapter, we presented the database structures that are key elements that need to be multiplexed and backed up to ensure a smooth recovery in the event of a database failure: the control files, online redo log files, and archived redo log files. The control files contain the metadata about every other structure in the database. The online redo log files provide performance benefits to ongoing transactions and ensure that no committed transactions are lost after an instance failure; being able to change the number of redo log groups and the number of members in each group enhances both the availability and performance of the database. Archived redo log files make copies of online redo log files to one or more destinations before they are overwritten by new transactions. The common thread through all three of these structures is multiplexing: creating redundant copies of database components or redundant archival locations to minimize the impact of a media failure.

We also introduced the Flash Recovery area and explained how it is the central location for backups of all database files, control files, initialization parameter files, and archived redo log files in the database. You can manage the Flash Recovery area via the EM Database Control interface or using a SQL command-line interface to set or change database initialization parameters that control its location and size.

Before making database backups, you must understand backup strategies, types, and modes. ARCHIVELOG mode provides many benefits and few downsides, especially in a production environment; NOARCHIVELOG mode, in many ways, restricts the types of backups you can make.

Recovery Manager, or RMAN, provides a number of benefits over manual backup methods using a combination of SQL and operating system commands. You can access most RMAN functionality via the EM Database Control or with a command-line version for advanced backup and recovery techniques. One of RMAN's many benefits is the ability to create compressed backup sets, which not only skips unused blocks in database datafiles, but compresses the blocks before writing to the backup set, saving I/O bandwidth and disk space.