Enhancing hosting infrastructure management with AI-powered automation

Abstract

Managing hosting infrastructure presents increasing challenges due to demands for high availability, efficient resource allocation, and strict security compliance, while traditional manual administration methods often result in inefficiencies and delayed incident responses. This thesis explores the potential of AI-driven automation, focusing specifically on integrating predictive analytics, real-time anomaly detection, and intelligent incident response mechanisms to enhance operational efficiency and resilience. Central to the research is the application of Retrieval-Augmented Generation (RAG), ensuring AI workflows adhere strictly to operational policies and minimize unpredictability. Utilizing AI agents for proactive monitoring, automated decision-making, and failure prevention, the study evaluates effectiveness within Proxmox VE and Ceph-based virtualized environments, emphasizing high availability clusters and predictive resource scaling. Findings demonstrate that AI-powered automation significantly reduces downtime, improves adaptability, and strengthens security through proactive incident management, dynamic workload distribution, and automated compliance enforcement, ultimately reducing administrative overhead and optimizing resource utilization.