Sarcouncil Journal of Engineering and Computer Sciences

Abstract: The transformation in cloud-native application design has further emphasized the necessity for smarter virtual machines and autoscaling systems in containerized environments. While autoscaling for stateless services has been established as a viable design, the use of more stateful applications introduces specific challenges related to session integrity, data consistency, and service availability during scaling operations. This paper describes a stateful autoscaling reference architecture that integrates proactive and reactive models, an SLA-aware algorithm, lifecycle management, and live migration strategies across multi-cluster environments. The architecture considers the operational semantics of containers and virtual machines and leverages orchestration tools such as Kubernetes and Docker to ensure high availability and efficient resource utilization. The proposed solution addresses the scalability limitations of existing autoscaling models by integrating principles of state management, predictive analytics, and orchestration—thus overcoming prior constraints. This architecture enables robust and flexible scaling of modern applications deployed in complex distributed systems.