With PARALLEL, each worker process will access one file at a time. So, assuming your LOB table is larger than 10G, and LOB accesses are always slow, I hazard a guess that gzip started after file reached about 10g, but before next set of rows fetched by worker process. The new file creation happens *when* worker process tries to write and finds that you have reached max limit and then it will create a new file and continue. if I were to gzip, I'd wait until next file is created and only then start gzipping. "ORA-08180: no snapshot", have you looked at this error yet? you are specifying flashback_time, so is undo_retention large enough to account for all the time since flashback time specified and current expdp time? LOBS normally take too long, so perhaps you could try adding access_method=external_table might help, YMMV. Also i didn't know FILESIZE accepted 10GB, I always thought it should be 10G, i.e. sans the B, something new to know. Raj -- //www.freelists.org/webpage/oracle-l