On May 4, 2:54 pm, joel garry <joel-ga..._at_home.com> wrote:
> On May 4, 6:52 am, EscVector <J..._at_webthere.com> wrote:
>
> > jg,
> > I'll post the stats. Still setting up.
> > Platform wise it has to run on any platform. I'll test on Win2003,
> > XP, AIX, Linux, Solaris and provide disk and setup specs.
>
> Ah, now that defines the problem much more clearly. If you had
> limited to unix/linux, there is a much less of a bottleneck writing to
> files, as well as shell tools always being available.
>
>
>
> > I do like the sqlplus option and it is quite fast in my unmeasured
> > alpha tests, but I want to the possibly to fork out process and pseudo
> > parallelize the disk writes, stop and start the process, etc. I can
> > do this with SQLPLUS but I'm not a huge fan of calling an .exe when I
> > need programmatic control. This bias might be more personal than
> > technical, but I've seen lots of so called batch jobs that call
> > sqlplus and simply ignore exceptions and control issues. I prefer
> > event driven notification via a class object rather than in-line batch
> > process with goto like exceptions. I realize the WHENEVER SQLERROR/
> > OSERROR has some level of control but the default is to exit and I've
> > had trouble getting large batch process to behave or continue
> > correctly when using this.
>
> Once you throw Win into the mix, your bias becomes much more sensible
> technically, especially the batch issue.
>
>
>
> > Ultimately I'm only providing test detail; others will decide what
> > they like best. I just need to provide the options to the available
> > skill-sets.
>
> Good luck!
>
> jg
> --
> @home.com is bogus.
> Happy to be absurdist.
jg,
thanks for the input.
Received on Mon May 07 2007 - 08:59:23 CDT
