Re: Loading 300 million records

I briefly read parts of the paper and do not find a single example which
might be faster than OP's original technique. Also, the paper seems to be doing "insert /*+ append nologging */" ... nologging is *not* a valid hint afaik. A table can be logging / nologging .. however one cannot make an individual insert statement as nologging.

I believe he needs to start with tuning the select statement and go forward from there (i.e. look at disk i/o). A 10046 trace would be helpful in identifying the wait times.

Anurag Received on Mon Jun 06 2005 - 13:14:39 CDT