Try to create your "extractmonth" table first, but using the NOLOGGING
option (same option for the indexes, if any)... I think this option
prevents the database from making any copy of the table in case of a
rollback (which takes a long time). But, note using this option, you
will not be able to rollback.
Jon Waterhouse wrote:
>
> I have a file with about 800,000 records from which I need to extract
> one month's worth of data (about 30,000) records.
>
> insert column_names into extractmonth
> (select column_names from fullfile where year = 1999 and month = 4);
>
> runs at about 2.5 hours.
>
> create table extractmonth as
> select column_names from fullfile where year = 1999 and month = 4;
>
> runs at about 2.5 seconds.
>
> There is an index on year and month on full file.
>
> Any ideas on what causes the difference?
Received on Tue Apr 02 2002 - 15:24:30 CST
