Software Architecture Evaluation Model for High Availability Cloud Applications

Abstract

High-availability cloud applications require architectures that can continue operating during service, node, zone, data, and regional failures. This study presents a software architecture evaluation model that assesses service redundancy, load distribution, fault isolation, data replication, and recovery automation before deployment. The model represents cloud components and dependencies, assigns availability risks, validates fault-tolerance rules, and recommends architecture improvements. It was evaluated across six increasing failure-intensity levels using four designs: a basic single-region architecture, a multi-zone architecture, a multi-region fault-tolerant architecture, and the architecture selected by the proposed model. Results showed that the proposed design maintained 99.18% availability at the highest failure level, compared with 87.40% for the single-region system. The model also improved failure containment, recovery speed, traffic redirection, and data continuity. The proposed approach can support cloud architects in selecting more resilient designs, reducing downtime, and improving service continuity for critical applications under demanding operating conditions and complex infrastructure disruptions.