IBM System/370 145 Manual page 57

processor storage required to operate the program concurrently
with the existing morning workload cannot be justified.
8. A terminal-based application has been installed on a full
production basis for sevel~al months. During this period, the
benefits accrued from the online application have encouraged the
gradual addition of severClll more terminals, and peak activity is
considerably higher than i t was initially. Because growth has
been gradual, much additiC::tnal programming time (significantly
more than is required to maintain batch-oriented applications)
has to be spent periodical.ly restructuring the terminal-based
application program to handle the increasing volume of activity.
9.
An online application is currently active during an entire shift
and operates concurrently with batched jobs. It would be
advantageous to install a second terminal-oriented application
that would operate concurrently with the existing workload during
the entire shift. However. the amount of processor storage that
would have to be dedicated to each online application for the
entire shift in order to handle its peak activity is very large,
and times of peak activity for the two applications do not
completely overlap. Because so much processor storage would be
unused during a large portion of the shift if both online
applications were always active, installation of the second
online application is difficult to justify.
In the situations described, processor storage is a constraining
factor in one way or another and ,the constraints highlighted can apply
in some degree to all systems regardless of their scale (small,
intermediate, large) or processor storage size. The fact that larger,
less expensive processor storage is now available on System/310 models
does not remove these constraints for two major reasons.
First, ,once a storage design point has been chosen for an
application, whether the design point is relatively large or small, the
application is dependent on that processor storage size for its
operation. The application cannoit. execute in less than its design point
storage amount, nor can it take advantage of additional available
processor storage without being ~)dified (unless i t has been
specifically structured to use additional storage as, for example, are
most IBM-supplied language translators). Second, although processor
storage has become less costly,
i1:
still is a resource that should be
used efficiently because of its importance in the total system
operation. Thus, when storage de;ign points are chosen, tradeoffs among
processor storage cost, application function, and system performance are
often made. Making applications fit within the storage design points
selected becomes the responsi'bilit:y of application designers and
programmers. This situation is made more difficult by the fact that for
many applications an optimum storage design pOint cannot be determined
until the application is written and tested using expected transaction
volumes.
The significance of processor storage restraints should be evaluated
in light of the following trends E~videnced by new types of applications:
(1)
the total amount of storage rE~quired to support their new facilities
continues to grow larger, (2) the storage they actually require for
operation during their execution is tending to become more variable, and
(3) it is becoming as desirable to install many of these new
applications on smaller scale syst.ems with relatively small maximum
processor storage sizes and low volume requirements as i t is to install
them on larger scale systems. Reduction of the constraining factors
currently imposed by processor storage is, therefore, a necessary step
in making new applications easier and less costly to install and
available to a wider range of data processing installations.
A Guide to the IBM System/310 Model 145 47