Re: Best Practises/parameters for speeding Datapump & exporting to Multiple directories

I don't think you will benefit by having PARALLEL > #CPU on your machine. For your case, you may specify more than 8, but some may idle. dumpfile syntax will work, but are you sure your file system can handle a single file worth 350g?

directory object is basically just a directory, it is nothing but an entry in the dictionary.

I am hoping table has a suitable PK exists on the table. So, I'd fire 4 (you may want to reserve some CPU for other tasks) parallel expdp sessions with QUERY clause.

I am still curious, but let me know if you manage to write a single 350g file.
YMMV
Raj

--
http://www.freelists.org/webpage/oracle-l