The increasing complexity of data management and storage systems, coupled with the growing demand for flexible and efficient solutions, has led to the emergence of data virtualization technologies. This study investigates the potential for enhancing data storage methodologies through virtualization approaches, particularly focusing on the integration of hierarchical storage systems. Additionally, the research explores methods for organizing the management of such systems. The principles, methodologies, and architectures underlying distributed storage systems employed for handling big data tasks are analyzed. Experimental validation was conducted through the physical implementation of a prototype system on hardware. The results demonstrate a hierarchical data storage system leveraging virtualization, facilitating seamless data access and integration from disparate sources independently of their structure or storage method. Furthermore, a management approach based on reinforcement learning is proposed for controlling the developed storage system.