Data Pump

This is one type of logical backup introduced in oracle 10g version.

Utilities of data pump are expdp and impdp.

Data pump is much faster compared to the traditional logical backup since it uses the API's like dbms_datapump and dbms_metadata.

Data pump is a server side utility which means it generates the dump file at the server side irrespective of the location from where we initiated the job.

The pre-requisite for data pump is we need to create a directory at oracle level as well as at operating system level.

By default, a user can take the backup of his own objects, incase of data pump user should have read,write privilege on the directory.

The biggest advantage of data pump is we can stop/detach the job whenever there is a performance issue.

Once we reinitiate the job, it starts from the point where it was stopped earlier(which means we can detach and attach a job).

Once we initiate the data pump job, a master control process gets invoked and it keeps track of the status of the job in a table created by the job name.

Recommanded to give a name for every job that we initiate with data pump.

If we dont specify the job name, oracle creates a table on its own naming convention.

Once the job is completed, table gets dropped automatically.

We can also take the backup parallally with the help of "parallel" parameter.

We can compress metadata as well as business data in the dump file with the help of "compress" paramerer.

We can also estimate the space required in the backup destination without actually taking the backup.

Data pump enables faster movement of data between the databases or across the databases.

Useful in case of database migrations and upgradations.

Also useful in case of schema level refreshes.

 

Note : If we give the same name for the dump file, exp overrides without any warning messages or error messages. expdp fails with error message "file already exists".

 

Note : We cannot take incremental backups using data pump.