Re: When to Commit?

Mark D Powell wrote:
> While long running jobs are often canidates for improvement I do not
> see hoow you can just assume that becasue a job runs for 36 hours that
> there is a design problem. I have seen tasks that ran close to the
> theoretical hardware performance limit for performing the work that ran
> more than 12 hours. Sometimes you just have that much data to process.
> You could argue whether the work needs to be done, but without knowing
> the process or the requirements for the task I do not see how you can
> make an intelligent determination based on run time alone.
>
> IMHO -- Mark D Powell --

Just for fun, if you don't mind, propose a "real" task that could conceivably run for 36 hours and the limiting resource that goes with it.

And I assume you mean a regular task ... not a one time data load.

Thanks.

-- 
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)