Without a careful approach to backing up your Oracle database, you run the risk of experiencing excessive system downtime and possibly losing data. The following graphic shows the main ways that data loss occurs:
How you react to data loss depends on how it was caused:
You must recover the database up to the point in time when the database crashed. If a full recovery is possible, only the data of uncommitted transactions before the error is lost.
You must recover the database up to a point in time shortly before the error occurred. However, data entered after the error is lost.
To avoid data loss after a logical error, it is sometimes possible to restore the database to a different machine and then export the affected table from that machine to your production database. However, this method is difficult and requires expert knowledge of the application that uses the table.
As well as regularly backing up the database, you also need to archive the redo log files. The following graphic shows how important it is to archive the redo log files:
We recommend you to keep at least two copies of the offline redo log files on a secure storage medium. For maximum security, store the copies in different locations.