Proactive Ai-Driven Framework For Disaster Recovery In Cloud Computing

In Cloud-dependent environments, DR is crucial for protection of data and continuity. This paper presents an AI-driven framework for DR in cloud computing, integrating anomaly detection, classification, and reinforcement learning. The Cloud Computing Performance Metrics Kaggle dataset was analyzed VM metrics (CPU, memory, network, power). The Isolation Forest algorithm detect anomalies from VM metrics with high accuracy and no false positives, improving risk assessment. A Random Forest classifier classifies VMs by health status with perfect performance metrics enabling interventions for specific failures. A Q-learning model optimized resource allocation of VMs at risk, reducing disruptions, and conserving system resources. The framework was tested on AWS with the use of CloudWatch for real-time metric monitoring and SNS for instant threshold alerts. This approach improves anomaly detection, classification, and recovery, with less downtime and more reliability in cloud environments.