-Real-time optimization of the overall performance of a large computer system or network such as an air-traffic control system or teleprocessing network inherently requires the introduction of adaptive control into selected control functions or sets of control functions. Such adaptive control is necessitated by the high variance of the available resource inventory and the workload resource demands within the system environment. Global management of the resulting multiloop control system becomes the responsibility of the operating system. Investigated aspects of this evolutionary extension of the operating system called the dynamically adaptive operating system (DAOS) include the general methodology and the real-time modeling and estimation of resource demands. The methodology specifies the general identification, decision, and modification processes required for an adaptive operating system. Support of such processes could be attractively provided by a peripheral or miniprocessor. The problem of estimating resource service times both with and without partial service times is explored and statistical models possessing utility for real-time modeling are developed. In particular, a dynamically partitioned second moment model (DPSMM) is described. The feasibility of such models is established with respect to a General Principle of Locality. Selected simulation results are presented to evaluate selected developments with respect to an adaptive central processing unit (CPU) scheduling discipline.

Aspects of a Dynamically Adaptive Operating System

Models constitute a useful means of investigating computer system performance. This paper examines the interrelationships between models and other methods for evaluating the performance of computer systems and establishes circumstances under which the use of a model is appropriate.

The role of computer system models in performance evaluation

The applicability and accuracy of a general queuing network approach to the modeling of computer performance was tested in an actual production environment. The validation exercise established that such an approach had both sufficient flexibility and precision to be of use in long range planning efforts. The accuracy of the results further suggests that these techniques may be successfully extended to a wide range of uses.

Validation of a computer performance model of the exponential queuing network family

The design, sizing and tuning of computer systems is a persistent, expensive, time-consuming, and difficult problem. An appropriate, integrated support methodology to support the needs of the computer center manager vis-a-vis these issues is lacking. Although the performance literature related to this topic is voluminous, much of it is directed to the vendor rather than the computer center manager.

A heuristic approach to computer systems performance improvement, I: a fast performance prediction tool

Beginning in mid-1974, the Data Management Center of the Office of the Secretary of the Department of Health, Education, and Welfare undertook to validate an analytic model of computer performance in the context of a large multiprogrammed configuration operating under a variety of actual production workload conditions. A primary goal of this validation effort was to test the adequacy of the framework offered by the “exponential queuing network” model schema (as developed by Buzen, et al.) in representing “real world” situations with sufficient accuracy to be useful as a general tool in support of long range planning. The purpose of this paper is to report on the results of that validation effort.

A probability-based, theoretical model of a multiprogrammed computing system is suggested for planning future computing center requirements.

Validation of the planning model is attempted with respect to the theoretical model and applications to short-range and long-range planning.

F. Hanssmann

Papers

Modeling for computing center planning