This article studies the effects of queueing delays, and users’ related costs, on the management and control of computing resources. It offers a methodology for setting price, utilization, and capacity, taking into account the value of users’ time, and it examines the implications of alternative control structures, determined by the financial responsibility assigned to the data processing manager.
--Author’s Abstract
This paper should be of value to the Data Processing (DP) manager seeking more (financial) control over the DP operation whether or not there is some form of chargeout system in place. If there is one, its value may be questioned by the author’s conclusions. If there is not one, the manager will be guided to an appropriate one by this paper. In addition the paper will be helpful in “proving” to management that the installation is operating at an efficiency level that is appropriate to its service requirements.
The author raises a number of issues and demonstrates some conditions that run counter to intuition and, in some cases, common computing practice. Among them are:
(1) When computing centers set prices to cover their costs, if delay costs are considered, such cost recovery would lead to under-utilization of resources. He suggests deficit pricing to encourage use and provide maximum organizational benefits.
(2) “Self-regulated” queueing or failure to take queueing effects into account is unworkable and leads to congestion. The solution is imposition of queueing-related charges.
(3) When computing centers are organized as profit centers there is a danger of monopoly pricing. With queueing delays the situation becomes extreme since the profit center arrangement often reduces available capacity and the relative utilization of this capacity.
(4) In defending the level of a computer center’s utilization the author’s results show that “taking users’ delay costs into account, seemingly low utilization ratios are often optimal.”
By his own admission, the author’s methodology is geared to business DP systems with regularly scheduled tasks and clear cost/benefit data. But he does have suggestions for handling situations with irregular demand, such as research and educational institutions and those situations faced with discontinuous costs or cyclical demand over time. An extensive appendix discusses the implications of Peak-Load Pricing to the author’s methodology.