Join this online group to communicate across IBM product users and experts by sharing advice and best practices with peers and staying up to date regarding product enhancements.
Oracle can do two types of backups - physical and logical. Physical backup is done via Oracle stand-alone tool RMAN - Recovery manager. In short. this backup backs up physical datafiles that make up the tablespaces, the objects owned by user which is equivalent to schema.
This type of backup is main backup Oracle uses to keep its database secure and to keep also previous versions of data to some agreed back in time level.
While physical backups are indispensable nowadays, the situation may change in near future, as moving to the cloud offers its own layer of protection where the database structures can be copied to multiple siloes simultaneously. The advantage of RMAN backups is also possibility to reverse to a certain SCN, SCN standing for system change number, virtually allowing to restore the physical layer of data to any state where any change occurred. RMAN does this using redologs storing all changes made to the database. Logical backup is done at schema level which is solely Oracle structure and OS systems cannot see those. It is done via export datapump utility, in brief expdp. This backup is very flexible as it backups up only precisely defined objects within schema=user ownership. They are also much simpler that physical backups, no need for TSM opt file and complex RMAN backup structures like backup pieces, a single command does the trick no matter how big the schema/s/ is/are. Logical backup is therefore essential for keeping the schema structure as of time when the backup is done. It can be stored then indefinitely or simply replaced by another backup. Another advantage is compression Oracle uses making possible ratios as big as 1/20. This is not possible with physical backup and can be considered as one of main advantages over physical backup. What is also possible is to make a physical backup called flashcopy. This kind of backup is done via copying the metadata - data about the data store in headers of each datablock that are many orders of magnitude smaller than raw data - we compare here bytes and terabytes and bigger units - first and then only raw data making availability of data on the other side instantly as metadata are copied usually in matters of seconds. All kind of backups mentioned can be automated, put into script and executed at regular time intervals.
Copy