Data Pump - Oracle RMAN for Absolute Beginners - page 234

You can manually apply the package's UNOBF function to see the real values of the column:

SQL> select obfus.unobf(last_name) from customers;
OBFUS.UNOBF(LAST_NAME)
-------------------------
Lopuz
Gennick
Kuhn

Suppressing a Log File
By default, Data Pump creates a log file when generating an export or an import. If you know that you don't want a log
file generated, you can suppress it by specifying the NOLOGFILE parameter. Here is an example:

$ expdp mv_maint/foo directory=dp_dir tables=inv nologfile=y

If you choose not to create a log file, Data Pump still displays status messages on the output device. In general, I
recommend that you create a log file with every Data Pump operation. This gives you an audit trail of your actions.

Using Parallelism
Use the PARALLEL parameter to parallelize a Data Pump job. For instance, if you know you have four CPUs on a box,
and you want to set the degree of parallelism to 4, use PARALLEL as follows:

$ expdp mv_maint/foo parallel=4 dumpfile=exp.dmp directory=dp_dir full=y

To take full advantage of the parallel feature, ensure that you specify multiple files when exporting. The following
example creates one file for each thread of parallelism:

$ expdp mv_maint/foo parallel=4 dumpfile=exp1.dmp,exp2.dmp,exp3.dmp,exp4.dmp

You can also use the %U substitution variable to instruct Data Pump to create dump files automatically to match
the degree of parallelism. The %U variable starts at the value 01 and increments as additional dump files are allocated.
This example uses the %U variable:

$ expdp mv_maint/foo parallel=4 dumpfile=exp%U.dmp

Now, say you need to import from the dump files created from an export. You can either individually specify the
dump files or, if the dump files were created with the %U variable, use that on import:

$ impdp mv_maint/foo parallel=4 dumpfile=exp%U.dmp

In the prior example the import process starts by looking for a file with the name exp01.dmp , then exp02.dmp ,
and so on.

■ Oracle recommends that the degree of parallelism not be set to more than two times the number of Cpus
available on the server.

Tip