ElAr :: Перегляд за автором "Zveriev, P. V."

Зараз показано 1 - 1 з 1

Adaptive methods for managing distributed computing with built-in selfhealing mechanisms
(NURE, 2025) Volk, M. O.; Buhrii, A. M.; Zaihraev, D. S.; Panchenko, Y. O.; Zveriev, P. V.
Cloud computing systems increasingly face challenges of heterogeneity, scalability, and reliability. This paper presents an integrated approach to distributed computing management with embedded self-recovery mechanisms in heterogeneous cloud environments. The architecture combines adaptive scheduling, predictive analytics, and machine learning methods (LSTM, Random Forest, reinforcement learning) to forecast failures and optimize recovery scenarios. Experimental evaluation demonstrates improved fault prediction accuracy (12–15%), reduction of false alarms (25–32%), faster incident response (from 11–17 minutes to 5–9), and a 17% decrease in operational costs. The results confirm the technical and economic feasibility of the proposed approach for missioncritical cloud infrastructures.