RE: Data Pump and compress on the fly in Unix

If datapump writes to its ouput files not just in a continuous append
stream, e.g. updating dumpfile headers after all's done, then a named pipe
cannot be used.
 
Tanel.

 
I'm not even convinced Oracle's software can tell the difference between a
stream of bytes from a named pipe and a stream of bytes from a file.

Other related posts: