Research on the efficiency of virtualization solutions for deploying digital twins in systems with limited resources

Abstract

The subject matter of the article is the means and approaches to virtualization in environments with limited resources, in particular on single-board computers, as well as their ability to ensure the effective operation of digital twins. The goal of the work is to evaluate the effectiveness of virtualization for deploying digital twins on single-board computers and develop recommendations for virtualization usage in environments with limited computing resources. The following tasks were solved in the article: analysis of modern hardware solutions for single-board computers and existing virtualization technologies; construction of an experimental environment for practical comparison of virtualization technologies on single-board computers; implementation of an application load that simulates the operation of a digital twin with real sensor data; study of the efficiency of network interaction between virtual environments and the host system; comparison of results by key metrics. The following methods used are: virtualization technologies Docker, QEMU, Firecracker; network monitoring; thorough analysis of literature and available documentation on the topic; comparative analysis of the achieved results. The following results were obtained: a comparative experiment using Docker, QEMU and Firecracker based on Raspberry Pi 4 was carried out; thorough analysis of literature and available documentation on the topic was conducted; a basic prototype of a digital twin with the connection of physical sensor devices that collect telemetry data in real time was implemented; phased automatic deployment of virtual environments with pre-configured components was provided for rapid scaling of the experiment; network delays, message processing time and platform resource usage were measured; the advantages and limitations of virtualization technologies in conditions of limited computing resources were determined; the practical feasibility of using each type of virtualization in conditions of limited computing resources was assessed. Conclusions: Docker provides the easiest deployment and highest efficiency, but has less isolation compared to QEMU and Firecracker; Firecracker demonstrates the optimal balance between performance, security, and resource consumption on single-board computers; QEMU has the highest overhead, and therefore the lowest efficiency; the use of lightweight hypervisors is advisable in digital twin systems on peripheral devices.