Author(s): Lucas Vermeer and Sanne de Vries

Abstract: Cloud computing has become a foundational infrastructure for teaching, experimentation, and research in academic institutions, where cost efficiency, performance predictability, and ease of management are critical constraints. Virtual machines and container-based virtualization are the two dominant deployment paradigms used in academic cloud setups, yet their performance trade-offs are often evaluated using assumptions derived from enterprise environments rather than educational contexts. This research evaluates computational, memory, storage, and network performance differences between virtual machines and containers when deployed in small to medium academic cloud environments with limited hardware resources. Benchmark-driven experiments were conducted using representative workloads commonly found in teaching laboratories and student projects, including web services, data processing tasks, and parallel computation exercises. Performance metrics such as startup latency, resource utilization, throughput, and execution overhead were systematically measured and compared across both virtualization approaches. The findings indicate that containers consistently demonstrate lower startup times and reduced overhead for CPU and memory intensive tasks, while virtual machines provide stronger isolation and more predictable performance under mixed workloads. Storage and network performance showed smaller differences, with configuration choices playing a significant role in observed outcomes. The results highlight that the perceived superiority of one technology over the other depends strongly on workload characteristics, administrative objectives, and pedagogical requirements. By contextualizing virtualization performance within academic cloud environments, this research provides practical insights for educators and system administrators seeking to balance efficiency, reliability, and instructional flexibility. The outcomes support informed decision-making regarding infrastructure design for academic clouds and suggest that hybrid deployment models can effectively leverage the complementary strengths of virtual machines and containers. Such evidence-based guidance is particularly valuable for institutions aiming to modernize curricula while maintaining operational simplicity, minimizing costs, and ensuring that students gain realistic exposure to contemporary cloud technologies through hands-on experimentation in diverse instructional and research scenarios globally today.

DOI: 10.33545/27076636.2026.v7.i1a.147

Pages: 11-15 | Views: 182 | Downloads: 130

Download Full Article: Click Here