Savchenko A. Methods of distributed management of corporate computer fences

Represented work has solved an important scientific and applied problem in the field of information technology based on the performed theoretical and experimental research. We had increased the efficiency of corporate computer networks in terms of increasing usable throughput with physical constraints on network resources in computer systems with non-stationary traffic flows. The study of the current state of corporate computer network (CCN) management in non-stationary flow of requirements allowed to select the main areas of research, such as the adequacy of analysis and forecast of information on the state of the network, delivery delays of signal and control information, optimality of produced control impacts with minimal resource costs. The improved method of analysis of statistical characteristics and flow models of heterogeneous network traffic requirements with marking (prioritization) of traffic flows, aggregated and sparse flows allows to obtain estimates of the predicted network load and adjust the parameters of controllers. A method of optimal control of a computer network has been developed, which, due to the application of a modified criterion of generalized work of Alexander A. Krasovsky using the informational criterion function, makes it possible to find optimal controls for a composite network in real time. The developed method of system stability control allows ensuring a stable state of the system with random delays of signal and control information. The mathematical model was improved in the form of the transition function of the network node as a controlled object. This model, in contrast to the known ones, takes into account the presence of delays in signal and control information. It provides optimal choice of response time constant of the controlled object and sufficient stability resource of the system. The improved method of signal and control information transfer, due to the definition of the optimal distributed hierarchical control structure and the implementation of control information transport in the autonomous network segment at the data link layer reduces the delivery time of service information by 25%. The developed method of evaluating the effectiveness of the network management system allows evaluating the quality of the control system on its impact on network performance. Developed information technology for corporate computer network management makes it possible to achieve quality of service (QoS) objectives for various services at minimal cost of information resources. The practical significance of the obtained results is determined by the fact that the proposed models and methods are the scientific and methodological basis for the development of information technology for CCN control systems using new methods of distributed control of available network resources in terms of signal and control information delays. The developed methods and information technology are implemented by integrating them with the corporate network management system and allow optimal redistribution of switching equipment resources to ensure QoS of different types of network services. The first section reviews and analyzes the current state of the problem of CCN management in non-stationary request flows, existing approaches, methods, hardware and software to solve the tasks. The second section discusses the features of modern traffic in corporate computer networks, identifies four main groups of flows and presents mathematical models for each type of flow: flow from an independent source, aggregated, sparse and labeled flows. The results of experimental traffic analysis are presented in third section and the method of analysis of statistical characteristics and flow patterns of modern CCM was improved. The fourth section develops a method of optimal computer network management based on vector functional with informational criterion function. The fifth section analyzes the stability of the state of the autonomous segment with delays in signal and control information and develops a method for control stability of the system based on forced return in the field of stability. The sixth section presents a mathematical model in the form of the transfer function of the router as a controlled object and proposes an approach to the choice of time constants of the switching node based on the results of asymptotic estimation of signal information delays. The seventh section develops the method for estimation the efficiency of control and develops information technology of management of CCN with stages of identification, forecasting, selection of optimal management strategy, which allows achieving QoS objectives for different services with minimal information resources required for this.