RE: ALTER TABLE ADD columns in parallel?

> THEN you might swap in place a one row table for the one having columns
> added and then manually follow Mark's suggestion after the patch finishes

This gave me an idea. Our upgrade involves an out-of-place data migration. Based on the SQL I saw being run -- i.e. no trace done yet -- I should be able to impdp all 3000+ tables except the data from this particular table. JDE can do what it feels it needs to do to it for the migration, then I can run an impdp of the lone table with a DOP of 16, create indexes, and reset DOP. Initial test shows minutes instead of hours, even disregarding obvious SAN caching (2-3GB/s on PIOs).

Of course, that's provided the migration doesn't need to apply DML to those 16M rows that I missed, but it's something to consider.

Thanks much!

Rich

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