INFORMATION TECHNOLOGIES FOR REAL-TIME MONITORING OF HETEROGENEOUS NETWORKS

Abstract

The article investigates the theoretical foundations and practical implementation aspects of information technologies for monitoring heterogeneous networks in real time. With the rapid development of digital infrastructure and the widespread integration of diverse devices and protocols within a unified information environment, the demand for high-performance and adaptive monitoring systems capable of operating effectively under dynamic network conditions is significantly increasing. The focus is on identifying contradictions between the high variability of traffic formats, frequent topological changes, and the requirements for data integrity and timely decision-making. It is established that traditional monitoring approaches, particularly those based on periodic polling or fixed tracking parameters, lack the flexibility necessary for responsive incident detection and network configuration updates in real time. The study proposes a formalized algorithm for processing data streams, incorporating transition conditions between the stages of anomaly detection, filtering, event correlation, and response actions. The analysis is based on widely used monitoring systems — Prometheus, Zabbix, Nagios, Datadog, and PRTG — for which a comparative model is constructed using metrics such as response time, detection accuracy, adaptability to topological changes, scalability, and resource consumption. The results justify the effectiveness of combining several technologies to compensate for individual weaknesses. The mathematical model includes metrics for network load, node failure probability, and processing time functions, enabling the implementation of an efficient adaptive response logic. For the first time, a structural-algorithmic scheme is introduced that enables cyclic analysis of current data and returns to the previous processing stage if network stability thresholds are not met. Future research perspectives involve integrating intelligent traffic classification methods and incident prediction based on neural networks, providing full automation in managing heterogeneous networks.