Multi-Cloud Management Architecture Design and Disaster Recovery Strategy for High Availability

Abstract

With the growing adoption of multi-cloud computing for mission-critical applications, ensuring high availability and designing resilient disaster recovery strategies have become paramount. Existing solutions often lack a unified architecture for managing heterogeneous resources and rely on reactive, rule-based recovery mechanisms. To address these gaps, this paper proposes the High-Availability Multi-cloud Management Model with Intelligent Orchestration and Disaster Recovery (HMM-IODR). The model integrates cross-cloud resource orchestration, real-time fault detection, automated failover, and consistent data replication mechanisms.A virtual multi-cloud testbed was implemented to emulate regional failures across provider-specific scenarios, with training and evaluation data derived from public M-Lab NDT measurements. Several disaster recovery strategies were evaluated, including multi-live deployment, hot standby, cold standby, and intelligent CDN-based traffic redirection. Experimental results showed that the multi-live deployment strategy yielded a Recovery Time Objective (RTO) of 3.45 seconds and maintained latency below 188.7 ms. This RTO represents an 88.5% reduction compared to the Cold Standby strategy under identical test conditions, demonstrating effective scalability. Further analysis of scalability metrics confirmed the effectiveness of the HADRM model in mitigating fault impact and enhancing overall system availability. This study provides a practical framework for building scalable, secure, and fault-tolerant multi-cloud information systems, contributing to the advancement of cloud-native distributed architectures.