Oracle data protection with ONTAP
NetApp knows the most mission-critical data is found in databases.
An enterprise cannot operate without access to its data, and sometimes, the data defines the business. This data must be protected; however, data protection is more than just ensuring a usable backup—it is about performing the backups quickly and reliably in addition to storing them safely.
The other side of data protection is data recovery. When data is inaccessible, the enterprise is affected and might be inoperative until data is restored. This process must be fast and reliable. Finally, most databases must be protected against disasters, which means maintaining a replica of the database. The replica must be sufficiently up to date. It must also be quick and simple to make the replica a fully operational database.
This documentation replaces previously published technical report TR-4591: Oracle data protection: Backup, recovery, and replication. |
Planning
The right enterprise data protection architecture depends on the business requirements surrounding data retention, recoverability, and tolerance for disruption during various events.
For example, consider the number of applications, databases, and important datasets in scope. Building a backup strategy for a single dataset that ensures compliance with typical SLAs is fairly straightforward because there are not many objects to manage. As the number of datasets increases, monitoring becomes more complicated and administrators might be forced to spend an increasing amount of time addressing backup failures. As an environment reaches cloud and service provider scales, a wholly different approach is needed.
Dataset size also affects strategy. For example, many options exist for backup and recovery with a 100GB database because the data set is so small. Simply copying the data from backup media with traditional tools typically delivers a sufficient RTO for recovery. A 100TB database normally needs a completely different strategy unless the RTO allows for a multiday outage, in which case a traditional copy-based backup and recovery procedure might be acceptable.
Finally, there are factors outside the backup and recovery process itself. For example, are there databases supporting critical production activities, making recovery a rare event that is only performed by skilled DBAs? Alternatively, are the databases part of a large development environment in which recovery is a frequent occurrence and managed by a generalist IT team?