Original Export and Import and Data Pump Export and Import differ on where dump files and log files can be written to and read from because the original version is client-based and Data Pump is server-based.
Original Export and Import use the FILE and LOG parameters to specify dump file and log file names, respectively. These file names always refer to files local to the client system and they may also contain a path specification.
Data Pump Export and Import use the DUMPFILE and LOGFILE parameters to specify dump file and log file names, respectively. These file names always refer to files local to the server system and cannot contain any path information. Instead, a directory object is used to indirectly specify path information. The path value defined by the directory object must be accessible to the server. The directory object is specified for a Data Pump job through the DIRECTORY parameter. It is also possible to prepend a directory object to the file names passed to the DUMPFILE and LOGFILE parameters. For privileged users, Data Pump supports the use of a default directory object if one is not specified on the command line. This default directory object, DATA_PUMP_DIR, is set up at installation time.
If Data Pump legacy mode is enabled and the original Export FILE=filespec parameter and/or LOG=filespec parameter are present on the command line, then the following rules of precedence are used to determine a file's location:
If the FILE parameter and LOG parameter are both present on the command line, then the rules of precedence are applied separately to each parameter.
Also, when a mix of original Export/Import and Data Pump Export/Import parameters are used, separate rules apply to them. For example, suppose you have the following command:
expdp system FILE=/user/disk/foo.dmp LOGFILE=foo.log DIRECTORY=dpump_dir
The Data Pump legacy mode file management rules, as explained in this section, would apply to the FILE parameter. The normal (that is, non-legacy mode) Data Pump file management rules, as described in "Default Locations for Dump_ Log_ and SQL Files", would apply to the LOGFILE parameter.
If a path location is specified as part of the file specification, then Data Pump attempts to look for a directory object accessible to the schema executing the export job whose path location matches the path location of the file specification. If such a directory object cannot be found, then an error is returned. For example, assume that a server-based directory object named USER_DUMP_FILES has been defined with a path value of '/disk1/user1/dumpfiles/' and that read and write access to this directory object has been granted to the hr schema. The following command causes Data Pump to look for a server-based directory object whose path value contains '/disk1/user1/dumpfiles/' and to which the hr schema has been granted read and write access:
expdp hr FILE=/disk1/user1/dumpfiles/hrdata.dmp
In this case, Data Pump uses the directory object USER_DUMP_FILES. The path value, in this example '/disk1/user1/dumpfiles/', must refer to a path on the server system that is accessible to the Oracle Database.
If a path location is specified as part of the file specification, then any directory object provided using the DIRECTORY parameter is ignored. For example, if the following command is issued, then Data Pump does not use the DPUMP_DIR directory object for the file parameter, but instead looks for a server-based directory object whose path value contains '/disk1/user1/dumpfiles/' and to which the hr schema has been granted read and write access:
expdp hr FILE=/disk1/user1/dumpfiles/hrdata.dmp DIRECTORY=dpump_dir
If no path location is specified as part of the file specification, then the directory object named by the DIRECTORY parameter is used. For example, if the following command is issued, then Data Pump applies the path location defined for the DPUMP_DIR directory object to the hrdata.dmp file:
expdp hr FILE=hrdata.dmp DIRECTORY=dpump_dir
If no path location is specified as part of the file specification and no directory object is named by the DIRECTORY parameter, then Data Pump does the following, in the order shown:
Data Pump looks for the existence of a directory object of the form DATA_PUMP_DIR_schema_name, where schema_name is the schema that is executing the Data Pump job. For example, the following command would cause Data Pump to look for the existence of a server-based directory object named DATA_PUMP_DIR_HR:
expdp hr FILE=hrdata.dmp
The hr schema also must have been granted read and write access to this directory object. If such a directory object does not exist, then the process moves to step b.
Data Pump looks for the existence of the client-based environment variable DATA_PUMP_DIR. For instance, assume that a server-based directory object named DUMP_FILES1 has been defined and the hr schema has been granted read and write access to it. Then on the client system, the environment variable DATA_PUMP_DIR can be set to point to DUMP_FILES1 as follows:
setenv DATA_PUMP_DIR DUMP_FILES1 expdp hr FILE=hrdata.dmp
Data Pump then uses the served-based directory object DUMP_FILES1 for the hrdata.dmp file.
If a client-based environment variable DATA_PUMP_DIR does not exist, then the process moves to step c.
If the schema that is executing the Data Pump job has DBA privileges, then the default Data Pump directory object, DATA_PUMP_DIR, is used. This default directory object is established at installation time. For example, the following command causes Data Pump to attempt to use the default DATA_PUMP_DIR directory object, assuming that system has DBA privileges:
expdp system FILE=hrdata.dmp
"Default Locations for Dump_ Log_ and SQL Files" for information about Data Pump file management rules of precedence under normal Data Pump conditions (that is, non-legacy mode)