Models for availability evaluation of file servers in private clouds

The demand for computing resources, including storage and processing capabilities, has surged in recent years. For example, during the COVID-19 pandemic, many activities typically conducted using local resources transitioned to online platforms. However, post-pandemic, services that were previously offered for free or at a reduced cost have become more expensive for companies or even for educational institutes. Consequently, it is crucial to analyze alternative solutions in terms of quality and services provided. This study conducts an availability analysis of Nextcloud, a file system hosted in a private cloud, through stochastic Petri net (SPN) and reliability block diagram (RBD) models. These models were validated via measurements performed on a real testbed cloud service, ensuring trust in the achieved results. Experiments were utilized to demonstrate the practical applicability of these models in real-world configurations. The outcomes of these studies revealed a significant reduction of nearly 85% in downtime compared to the base architecture, underscoring the effectiveness of the proposed approach in enhancing system availability.