Starting Oracle 10g, the old export/import tools were replaced by a newer, more sophisticated tool: the data pump.
This new tool had some advantages over the older tool – it was able to run in parallel, it has a better interface and it can be ran from PL/SQL code easily. It can even be restarted or resumed if needed. On the con side, it uses the database DIRECTORY object which means the file we’re exporting to or importing from must reside on the server.
The problem start when we try to diagnose what is going on when there is a performance issue. Since the process we use for exporting might be forked into multiple processes, we can’t really know what is going on behind the scenes.
So, how can we resolve that?