Progettazione e implementazione di un Cluster Kubernetes on-premis

This paper describes the design and implementation of an on-premise Kubernetes Cluster infrastructure at the National Institute of Geophysics and Volcanology (INGV) in Rome. The project was born from the need to modernize the existing infrastructure, based on a monolithic WebFarm, which showed increasing limitations in terms of scalability, security and IT resource management. The previous architecture required continuous manual interventions, was inefficient in managing IP addresses and vulnerable to frequent cyber-attacks. The main objective was to migrate applications and web services to a new cluster architecture that ensures automated management, greater reliability and operational flexibility. The Kubernetes system was chosen for its container orchestration capabilities, improving both horizontal and vertical scalability, as well as increasing security and optimizing resource use, it is able to detect and manage failures autonomously. If a container fails or becomes unresponsive, Kubernetes will automatically replace it, ensuring continuous availability of the application. The system also manages load balancing between service instances efficiently. A comparison between different deployment architectures is also presented, and the solutions implemented for Kubernetes cluster management and application protection are analyzed. A Continuous Delivery system (ArgoCD) has been implemented that automates the deployment process, reducing manual work and improving consistency between environments. Applications are automatically deployed when a change is detected in the Git repository. In this mode, each update is applied in a controlled and repeatable manner. Finally, future developments are discussed, including the federation of multiple geographically distributed clusters to further increase the resilience and availability of services.