<p>Data availability is critical in cloud environments, yet hardware failures and resource stress pose significant risks. Data loss can occur for a variety of reasons, including hardware failure or resource stress situations. To address this challenge, this paper presents a proactive virtual machine (VM) management system for OpenNebula cloud platform. This system provides instant recovery from service failover, ensures high availability, optimizes resource utilization, and reduces downtime in dynamic cloud environments. Our system leverages OpenNebula API as a prototype to dynamically implement and adjust replication limits based on real-time monitoring of virtual machine and host resource utilization, especially CPU utilization. The system combines replication VMs migration and VMs cloning techniques to provide a flexible and robust solution. Preliminary experiments demonstrate that replication offers improvements compared to cloning. Both mechanisms provide enhancements compared to taking no action when detecting a resource-constrained situation. Experiments show it reduces active hosts by 10%, achieves replication efficiencies of 65–82%, and enhances resource utilization compared to reactive approaches.</p>