Re: Sql*load with named piped giving no performance increase. What's wrong?

daven <davenport_at_hotmail.com> wrote:
>
>we have a unix shell script that does 2 processes, a cut command of a file
>(takes out a chuck of a file) and then a execution of sql*loader program that
>loads the result of that cut.
>
>With out a named pipe process takes ~20 minutes, the cut runs first then the
>sqlloader program starts after cut finishes
>
>we create a named pipe and then have the cut program feed the sqlloader
>program we notice that the sqlloader program starts up right away. however
>the total time is still ~20 are we doing anything worng?

Investigate the SQL*Loader control file syntax in more detail - I used to to something like this until I discovered that the control file syntax is powerful enough to deal with simple cases like this. I'd guess that you'd be able to do away with the preprocessing step altogether.

-- 
Andrew Mobbs - http://www.chiark.greenend.org.uk/~andrewm/