Re: Inserting 300.000.000 datasets ..

I'm a little out of practice with SQL*Loader, but named pipes should work just fine on UNIX.
The process would be to create the named pipe. Then, with a background job start writing the data into the pipe. This must be continuous to avoid an EOF. Use the pipe name from the first step as the input file for SQL*Loader.

I am not aware of any way to use stdin to get the data into SQL*Loader. The data can be inline with the control information.

1. create named pipe - mknod pipe_name.dat p
2. write the datasets to the pipe with a background job. This may need to be a custom job. - build_my_datasets > pipe_name.dat &
3. Run SQL*Loader - sqlldr userid=scott/tiger control=xxxxxx.ctl log=xxxxxx.log data=pipe_name.dat

HTH
   James

In article <39743252.2177372_at_news-server>,   nsouto_at_nsw.bigpond.net.au.nospam (Nuno Souto) wrote:
> On Mon, 17 Jul 2000 18:46:22 GMT, "Hans-Jürgen Schönig"
> <a9601497_at_unet.univie.ac.at> wrote:
>
> >
> >How does it work using STDIN with SQL*Loader? can I simply write:
> >xy | sqlldr db/db - ??? I've tried something like this but it didn't
 work.
> >
>
> Sorry, I can't make it work either. I remember vaguely that in V6 it
> was possible to have the control statements and the data coming in
> from STDIN. But I've just poured over all the V8 and 8i doco and can't
> find how to do it now.
>
> Does anybody else know how?
>
> Without that, I'm afraid a named pipe won't work.
>
> Cheers
> Nuno Souto
> nsouto_at_nsw.bigpond.net.au.nospam
> http://www.users.bigpond.net.au/the_Den/index.html
>

--
Life is complex; it has real and imaginary parts.


Sent via Deja.com http://www.deja.com/
Before you buy.