ARCHITECTURAL TRADE-OFFS BETWEEN STATELESS AND STATEFUL MICROSERVICES IN LARGE-SCALE CLOUD SYSTEMS

Abstract

Through a controlled experimental review, this study examined the architectural trade-offs of stateless and stateful microservices in large-scale cloud systems. To enable a fair comparison, Kubernetes was used to implement parallel installations of both designs on AWS with similar hardware specs. Three scenarios—normal load, peak load, and failure conditions—were used to evaluate performance, scalability, fault tolerance, and operational overhead. The findings showed that, especially in situations with fault injection and high traffic, stateless microservices produced higher throughput, quicker recovery times, and reduced operational complexity. On the other hand, stateful microservices performed exceptionally well in terms of transactional consistency and session retention, albeit at the expense of slower scalability and more maintenance work. According to the results, a hybrid architectural strategy that combines externalized state management with stateless service layers may offer large-scale cloud deployments the best possible balance between data integrity, flexibility, and resilience.