Towards Self-Healing Cloud Infrastructure: Automated Recovery Methods and Their Effectiveness

This study analyzes existing strategies for automated recovery within self-healing cloud infrastructures. The research is grounded in a review of findings from previous scientific publications. The analysis demonstrates that intelligent remediation methods can not only reduce downtime but also enhance the economic resilience of cloud infrastructure, paving the way toward fully autonomous, self-healing digital platforms. The scientific contribution of this work lies in the first comparative evaluation of the effectiveness of rule-based approaches, ML-prioritized methods, genetic algorithms, and DQN agents in multi-cloud Kubernetes environments. Its practical significance is reflected in the proposed modern approach of implementing a hybrid pipeline with a DQN-based scheduler, which achieves more than a 70% reduction in downtime and establishes a balance between recovery speed and cost-efficiency in real-world cloud platforms. The insights presented in this study will be particularly valuable to researchers in the field of autonomous distributed systems and cloud infrastructure reliability, especially those engaged in the development and formal verification of self-healing and automated failure correction mechanisms. Furthermore, the analysis of the effectiveness of these techniques holds practical relevance for leading DevOps/PlatformOps architects and SRE specialists seeking to enhance the availability and resilience of critical services through the integration of advanced automated recovery algorithms.