| Previous | Table of Contents | Next |
by Dan Hotka
Any computing environment is subject to a variety of hardware or software failures and problems that can cause a loss of stored data. Natural disasters, power outages, computer program problems, and machine failures are hard to predict. Other considerations that need to be taken into account are accidental and intentional corruption or destruction of data.
Each application has specific data requirement needs. Some applications are data-entry-oriented (merchandise sales via a cash register or a terminal) where others are just information retrieval (typical data warehouse). In the event of software/hardware failure, for whatever reason, the ability to restore the data after the failure is corrected will be necessary. A backup in a computing environment entails making an electronic copy of data and/or programs on removable media. The frequency and extent of backups depend on the application and business needs. For example, a database that is loaded from flat files every evening may not require that any backups be performed at all, whereas OLTP applications may not have any acceptable data loss or downtime.
A system-level backup will copy all changes in the system configuration, programs, and data to a removable medium such as tape or a storage management device. An Oracle database backup is to recover from any type of error, with minimal data loss and a minimal amount of application unavailability. Oracle database backups have multiple options for the many business requirements for the applications and data. These various backup/business scenarios is the topic of this chapter.
These availability and backup needs can be met with both hardware and software solutions. Hardware solutions include RAID (with its disk mirroring options), uninterrupted power supply (enough power to bring the machine down gracefully or to maintain full hardware availability for hours or even days), and redundant standby systems. Software solutions include software mirroring of files (Oracle supports mirrored redo logs and control files), software to maintain the redundant standby system (Oracle’s archive log mode and standby database options), and backup/recovery software (Oracle supplied, hardware vendor supplied, and third-party packaged software).
Applications address business needs. Availability, acceptable downtime, and acceptable data loss are all critical requirements when arriving at backup and recovery solutions.
The data-accessibility requirements need to be completely understood prior to the production status of any application. The type of backup and recovery plan to implement depends on issues such as acceptable downtime, recovery time, intervals of backups, and acceptable data loss. Acceptable data loss and acceptable downtime will determine the type of backup strategy required for any application. The more critical the information and consistent availability, the more redundancy in both hardware and software will be required. Many hardware platforms support a great deal of fault tolerance, and some vendors specialize in fault tolerance. Oracle has several mirroring features that enhance the computer hardware fault tolerance and data recoverability.
Many business needs dictate minimal or absolutely no application downtime. Any data lost in a retail sales application would mean a loss of sales. Any loss of collected credit card charge data would be a loss in revenue. Any data lost in an emergency application could mean a loss of life.
Both hardware and software features need to be considered by the business need. Redundant and fault-tolerant hardware systems can provide fail-over systems, disk mirroring, and uninterrupted power. Oracle supports many features that aid in the protection of data. Oracle supports instance level recovery via the redo logs. In the event of media failure, Oracle’s ARCHIVELOG mode can recover data to the last committed transaction. Oracle supports the software mirroring, or duplexing, of many critical files, such as the redo logs and control files. Oracle supports the concept of a “standby database,” where two identical Oracle environments are maintained, one in active use and one in constant recovery mode. ARCHIVELOG mode is essential in this configuration because as the archive logs are created on the primary system, they are moved and applied to the standby system. In the event of machine failure, the standby database can be quickly brought online and is as current as the last applied archive log. The amount of acceptable data loss is in inverse proportion to the frequency these archive logs are created and applied to the standby database environment.
| Previous | Table of Contents | Next |