🤖 AI Summary
Microservice systems commonly exhibit resilience deficiencies—including localized fault propagation, cascading timeouts, and inconsistent recovery behaviors—yet existing research remains largely descriptive, lacking systematic evidence synthesis and quantitative evaluation. To address this gap, we conduct the first PRISMA-guided systematic literature review (SLR) of 26 high-quality empirical studies published between 2014 and 2025. Our analysis identifies nine core recovery patterns and introduces three novel, empirically grounded artifacts: (1) a reproducible Recovery Pattern Taxonomy; (2) a standardized Resilience Evaluation Score for quantitative assessment; and (3) a constraint-aware decision matrix that explicitly trades off latency, consistency, and cost. Collectively, these contributions establish a structured, quantifiable, and reproducible empirical foundation for resilience-aware microservice design and engineering.
📝 Abstract
Microservice based systems underpin modern distributed computing environments but remain vulnerable to partial failures, cascading timeouts, and inconsistent recovery behavior. Although numerous resilience and recovery patterns have been proposed, existing surveys are largely descriptive and lack systematic evidence synthesis or quantitative rigor. This paper presents a PRISMA aligned systematic literature review of empirical studies on microservice recovery strategies published between 2014 and 2025 across IEEE Xplore, ACM Digital Library, and Scopus. From an initial corpus of 412 records, 26 high quality studies were selected using transparent inclusion, exclusion, and quality assessment criteria. The review identifies nine recurring resilience themes encompassing circuit breakers, retries with jitter and budgets, sagas with compensation, idempotency, bulkheads, adaptive backpressure, observability, and chaos validation. As a data oriented contribution, the paper introduces a Recovery Pattern Taxonomy, a Resilience Evaluation Score checklist for standardized benchmarking, and a constraint aware decision matrix mapping latency, consistency, and cost trade offs to appropriate recovery mechanisms. The results consolidate fragmented resilience research into a structured and analyzable evidence base that supports reproducible evaluation and informed design of fault tolerant and performance aware microservice systems.