A Comprehensive Taxonomy and Comparative Analysis of Fault Tolerance Mechanisms in Cloud-Native Microservice Architectures

Cloud-native microservice architectures have changed modern software systems but they also introduce distinctive and common reliability challenges as a result of extreme distribution, runtime dynamism and rapid change. Failures in such systems are hardly isolated; instead, they come from complex interactions between services, platforms and control policies which usually lead to cascading and metastable behaviours that conventional fault tolerance approaches fail to capture. Considering the volume of literature on resilience patterns, existing work remains fragmented, pattern-centric and weakly connected to observed failure models. This paper presented a systematic literature review and synthesis of fault tolerance mechanisms for cloud-native microservices by using recent peer-reviewed research and authoritative industry practice. The study constructed a multi-dimensional taxonomy that classified mechanisms in architectural layers, mechanism families, fault-handling phases, and runtime control characteristics by using a structured review methodology. A comparative matrix evaluated key mechanisms against operational criteria including latency overhead, scalability impact, complexity and risk of failure amplification. Building on this analysis, the paper mapped mechanisms to common cloud- native fault models and derived practitioner-oriented decision guidance. The results pointed out that resilience in cloud- native systems is dominated not by redundancy alone but by effective containment, observability and context-aware control. Misconfigured retries and static policies consistently amplify failures while adaptive and observability-driven approaches remain under-explored. The paper concluded by identifying concrete research gaps and testable hypotheses as well as providing both actionable design guidance and a foundation for future resilience engineering research.