Browse Prior Art Database

Windows for Automatic Processing in an Overloaded Environment

IP.com Disclosure Number: IPCOM000120892D
Original Publication Date: 1991-Jun-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 3 page(s) / 119K

Publishing Venue

IBM

Related People

Pence, JW: AUTHOR

Abstract

As computing requirements grow and automation is achieved, it becomes common that performance requirements for primary functions, such as direct customer service, cause secondary activities, such as housekeeping and clean-up, to be queued and performed as time and resources permit. Sometimes simple FIFO (first-in, first-out) queues are sufficient and create little difficulty for later processing, but in other cases, it is easier to record work-to-do in existing keyed data repositories. This latter situation frequently results in some challenging problems particularly when there is more queued work than can be accomplished in the available, bounded amount of time permitted to have it performed. This article provides an algorithm to 1. enable the fully allowed window to be used if there is work to be done, but yet 2.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Windows for Automatic Processing in an Overloaded Environment

      As computing requirements grow and automation is
achieved, it becomes common that performance requirements for primary
functions, such as direct customer service, cause secondary
activities, such as housekeeping and clean-up, to be queued and
performed as time and resources permit.  Sometimes simple FIFO
(first-in, first-out) queues are sufficient and create little
difficulty for later processing, but in other cases, it is easier to
record work-to-do in existing keyed data repositories.  This latter
situation frequently results in some challenging problems
particularly when there is more queued work than can be accomplished
in the available, bounded amount of time permitted to have it
performed.  This article provides an algorithm to
1.  enable the fully allowed window to be used if there is work to be
done, but yet
2.  perform the work in a manner that controls the maximum delay when
more work is requested than can be performed within the time
constraints.

      Another way to state the problem:  How to automatically handle
the environment where there is frequently greater than one hour's
worth of work requested but only an hour or less is provided each day
to perform it.

      A processing window will be defined.  The start can be caused
explicitly or be automated via a time-of-day function.  The window
may have a stop time or may not be bounded.  If there is no stop
time, then the process once started runs to completion.  If there is
a stop time, the function must check the time of day periodically and
stop if the time of day is beyond it.

      It is likely the case that these work requests which are
imbedded in an inventory are sparsely located and can be found only
by reading all the inventory.  Germane to performance of restarting
any such scanning function is the periodic recording on non-volatile
storage the position or key of the record just read.  The saved
record key can be provided to efficiently reposition should an
interruption occur.  The maximum amount of repeated scanning relates
to the frequency that you record the key during normal processing.

      Some applications consist of multiple functions each having
secondary work requests recorded in indexed inventories.  In these
cases the set of functions must be viewed as one with respect to
processing to the end and then if time permits, continue from the
beginning of the first function in the sequence.

      Following are the permutations of the conditions and the
resulting scanning domain when they exist.  The conditions are
whether there exists a saved key and/or a stop time.  The domain will
have a starting point, an ending point and possibly a discontinuity
in the middle.
           Condition            Scan Domain
         Sv-...