I'm in digest mode, so I apologize for wasting additional bandwidth on this
subject.
 
Gavin wrote:
>I got a call from a customer today who is having problems setting up
>a job which, among other things, :STREAMs another instance of itself
>to run the next day. Now you would think that this would be the most
>obvious use of the Scheduled Job options of the :STREAM command.
 
   As good as HP3000 systems are, it still amazes me that something so simple
can be utterly impossible to achieve. We had the same problem a few years ago
and our time drifted on us quite frequently. We actually had an IBM operator
that could restream the job manually for us. That in itself was a minor miracle.
We solved the problem as such:
  1.  We increased the amount of business so much that we required a multiple
processor  machine.
  2.  We created multiple 'groups' of our databases because the datasets were
too           large.
  3.  We ran our batch processing in parallel cycles, thus requiring parallel
job
       queues.
  4. Because of the complexity of running in parallel we were able to persuade
with        much begging and pleading the purchase of a $10,000 batch scheduler.
  5.  We can run our job at 6:14 each day by simply scheduling it at 6:14 each
day.        Amazing isn't it.
  6.  When things got really complicated and we brought in a second machine to
distribute processing, Gavin had to fix that very scheduler. It seems I can't
get        anything to be simple anymore.
 
Thanks for the use of bandwidth, I' feeling much better now. Product names are
withheld to protect the guilty.
 
Chad Gilles  [log in to unmask]
 
My opinions are mine and I refuse to let anyone have them!