The compression feature in 10g is related to the metadata, not the actual data part of the dump files. With 11g, this feature was improved to allow either the metadata, the row data or the complete dump file set to be compressed. This shrinks the dump file set by 10 to 15 percent.

In 11g it is possible to use the encrypted dump file sets feature to have the dump set encrypted. Data Pump in 11g includes other keywords to manage encryption, such as ENCRYPTION_ALGORITHM, and ENCRYPTION_MODE which requires the Transparent Data Encryption(TDE) feature to perform the encryption process. This feature will be addressed in more depth in the security chapter

In 10g, when a row in an external table was corrupted, it led to the entire process being aborted. Data Pump 11g is more tolerant under these circumstances, allowing the process to continue with the rest of the data.

With this feature, it is possible to move just one partition or sub partition between databases without having the need to move the whole table. A partition can be added as part of an existing table or as an independent table.

In 10g dump files had to be removed by the DBA prior to any attempt to overwrite them. In 11g a new keyword was added, REUSE_DUMPFILES, which defaults to 11g, and when activated simply overwrites the existing dump files (if they already exist).

In previous Data Pump releases, the XML data type was not supported, all Oracle XML data types are supported with Oracle Data Pump. You can use all other datatypes, however you should be aware that the Data Pump driver for external tables restricts the use of certain data types.