The material presents a real problem inherent in the management of computer systems, namely that of finding the appropriate system settings and thus being able to achieve the expected perfor- mance. The material also presents a prototype which aims to adapt the system in such a way as to achieve the objective, defined as the application efficiency. The prototype uses a resource-oriented mechanism that is built into the OS Workload Manager and is focused on a proposed goal-oriented subsystem based on fuzzy logic, managing resources to make the best use of them, and pursuing translation to the use of system resources, including nondeterministic technology-related factors such as duration of allocation and release of the resources, sharing the resources with the uncapped mode, and the errors of performance measurement.
The paper presents a new elastic scheduling task model which has been used in the uniprocessor node of a control measuring system. This model allows the selection of a new set of periods for the occurrence of tasks executed in the node of a system in the case when it is necessary to perform additional aperiodic tasks or there is a need to change the time parameters of existing tasks. Selection of periods is performed by heuristic algorithms. This paper presents the results of the experimental use of an elastic scheduling model with a GRASP heuristic algorithm.