Performance and Reliability Assessment of Cloud-Native Intelligent Systems

Abstract

Cloud-native intelligent systems increasingly support high-consequence decisions across operational, clinical, environmental, and industrial domains. While these systems promise elasticity, resilience, and continuous intelligence, their real-world effectiveness depends on measurable performance and sustained reliability under dynamic conditions. This paper presents a comprehensive assessment framework that evaluates latency, throughput, fault tolerance, prediction stability, and decision consistency in cloud-native intelligent systems. The framework integrates architectural analysis, uncertainty-aware analytics, and operational metrics to examine how system behavior evolves under workload variation, partial failure, and data uncertainty. Empirical results demonstrate that performance and reliability are not solely determined by infrastructure scalability, but also by model behavior, decision logic, and governance mechanisms embedded within the system.