Hassan A. Methods for estimating the operational efficiency and content of server computer systems

Object of research - a process server maintenance of computer systems based on the monitoring of their performance. The purpose- the creation of new ways of estimating the efficiency of the operation and content of server computer systems based on a common approach to the analysis of the characteristics of the server computer system based on neural network technology. Research methods - methods of the theory of computer networks and queuing theory of probability and statistical processing methods, artificial neural networks, fuzzy logic, metaheuristics, the methods of system analysis. Hardware - The server computer system. Theoretical and practical results - the algorithm of the proposed method for calculating the factors of server computer systems, which allows a high degree of accuracy to obtain a comprehensive assessment of the functioning; produced statistical estimates of the factors related to the type of web resources, which makes it possible to increase the accuracy of the content analysis; developed procedures that provide both adequate and learning models estimating the functioning and content. Novelty - first proposed a unified approach to the analysis of the characteristics of the server computer system based on artificial neural networks, fuzzy logic and metaheuristics, which solves the problem of increasing the efficiency of quality assessment and content for different architectures; further developed the method of calculating the characteristics of computer systems, which are characterized in that focused on the particular server computer system that improves the accuracy of estimation of efficiency of functioning; first created fuzzy neural network model estimating efficiency of the server computer system, which is characterized in that, using the formalization of knowledge in the form of fuzzy rules, a combination of factors makes it possible to give a comprehensive evaluation of the functioning, simplifying human interaction with the computer system; first developed a method for estimating the content server computer system, which is characterized in that it takes into account the statistical model is a neural network characteristics of the factors that can increase the probability estimates content of 15%; an improved method for determining the parameters of the proposed neural network models, which in contrast to the existing performance-based metaheuristics combination that provides both value and learning models. The degree of implementation - the results of the research are used in the Donetsk National Technical University in the design and maintenance of server computer systems. Use sphere - server computer systems.