Monitoring and Load Forecasting of IP Telephony Server in Digital Educational Ecosystems

Introduction. The article addresses the relevant scientific and practical problem of developing and substantiating a subsystem for intelligent server monitoring and load forecasting in IP telephony networks. It highlights the critical importance of transitioning from traditional reactive observation of telecommunication nodes to proactive infrastructure management. This is particularly vital for digital educational ecosystems, where the time series of telephone traffic exhibit a complex structure with pronounced daily and weekly seasonality, directly influenced by academic schedules, consultations, and examination periods. The purpose of the paper is to develop a software module utilizing the Holt–Winters method (triple exponential smoothing) for short-term traffic forecasting. Results. A detailed comparative analysis of existing monitoring tools (Zabbix, Prometheus, Homer) was conducted, identifying a significant gap between hardware resource tracking (CPU, RAM) and specific SIP signaling parameters. It was established that most systems focus on historical data analysis, lacking tools for predicting future peak loads. The scientific novelty lies in the experimental determination of the optimal model parameters (α = 0.2; β = 0.05; γ = 0.1), ensuring high stability against anomalous spikes and noise typical of VoIP networks. It is demonstrated that using this method achieves a MAPE accuracy of < 15 % without the need for resource-intensive neural network architectures. The practical value of the results involves integrating the developed algorithm into the modern VictoriaMetrics and Grafana technology stack. This solution automates the detection of overload risks and ensures the stability of communication services during peak periods of educational activity. Conclusions. The findings establish a methodological foundation for the further implementation of decision support systems within the IT infrastructure of educational institutions.