Re: UTL_FILE vs PRO*C/C++

Andy,

 Thank you for you input, I really appreciate! I thought that blaming UTL_FILE might be an easy solution but my query runs very fast and the bloody thing tales for ever to get on disk...

 From my point of view, it might be a nice solution for Oracle to send some consultants at a 1000$/day... But, hey, maybe I don't understand properly. B-))

 Thanks again!
 P

Andy Cowling wrote:
>
> >>>>> "Patrick" == <patrick.drouin_at_remove_this.nortel.com> writes:
>
> Patrick> I am using UTL_FILE routines to generate huge flat files
> Patrick> from the
> Patrick> content of the database. It turn out that the UTL_FILE
> Patrick> routines seem to be very slooowwww.
>
> I had an identical experience recently using Oracle 7.3.3.4 on
> Sequent/ptx 4.4.2. Unloading a 500k row table with c.20 columns took
> c. 40 mins.
>
> Patrick> I am thinking of writing these routines using PRO*C/C++
> Patrick> to speed things
> Patrick> up. Anyone ever went this way? I believe it might speed
> Patrick> things up quite a bit.
>
> I also went this way - Pro*C using host arrays (5k batches) improved
> the speed to < 2 mins ! Same source table, same m/c, same load, same
> disk. The improvment was so marked, I didn't even bother profiling the
> PL/SQL version further.
>
> Of course, if Oracle offered a product to unload tables into flat
> files (SQL*Unloader), we wouldn't all have to develop/maintain/profile
> N custom versions.
>
> I can only guess that Oracle Corp. refuse to offer such a tool as it
> could potentially be used as a migration (away from Oracle) utility as
> well as an invaluable DBA tool.
>
> Hope this helps.
> --
> Andy Cowling
> Sequent Computer Systems, Weybridge, UK
Received on Fri May 29 1998 - 00:00:00 CDT