Isn't the technical 'problem' being solved here *writing* a csv file with a header? So I don't believe external tables will help (because for some reason Oracle haven't implemented writing to csv in the external table arena).I'd be interested in where the file goes next - i.e the business problem. csv is usually a data transfer solution, to another db, to excel etc etc. There are likely better solutions for that that may not write a file at all. On May 30, 2012 5:17 PM, "rjamya" <rjamya@xxxxxxxxx> wrote: > I second Jared's recommendation for using external tables. you can use them > in parallel if large files can be split into multiple. Plus you can use the > magic of SQL to do necesasary transformation as well. > I have had great success with them. In once case the preferred ETL tool > couldn't do the job properly for a 6m-20m rows file in allotted time. We > managed to split the file into pieces, used parallel processing (one thread > for each file piece) for a single external table, and then use > dbms_errorlog to capture invalid data while loading into staging tables. > > Raj > > On Wed, May 30, 2012 at 11:44 AM, Jared Still <jkstill@xxxxxxxxx> wrote: > > > You may want to consider external tables and skip a lot of the code. It > > can be done much simpler with a SQL statement. > > > > > -- > //www.freelists.org/webpage/oracle-l > > > -- //www.freelists.org/webpage/oracle-l