Pemetaan Sistematis Kinerja dan Efisiensi Virtualisasi pada Infrastruktur Komputasi Awan

Abstract

Virtualization is a central foundation of cloud computing, yet the claim that it consistently improves efficiency with only minor performance degradation cannot be generalized without considering workload, platform, and configuration. This study maps evidence on performance, resource efficiency, and implementation conditions of virtualization in network and cloud infrastructures. A structured literature review was conducted on publications from 2010 to 2026 retrieved through IEEE Xplore, ACM Digital Library, ScienceDirect, SpringerLink, Google Scholar, and Indonesian journal portals. Eighteen core studies were retained based on relevance, measurable parameters, configuration clarity, and traceable findings; narrative synthesis was applied because hardware, operating systems, and test scenarios were heterogeneous. The findings show that virtualization overhead is not a fixed value. Performance is shaped by hypervisor type, storage and network paths, paravirtualized drivers, overcommitment, inter-virtual-machine interference, and guest operating-system compatibility. Proxmox VE, VMware ESXi, and Hyper-V may deliver comparable performance under selected workloads, but platform superiority changes across metrics, operating systems, and virtual-machine density. Containers generally provide lower overhead, although their isolation model is not sufficient for every workload. The study proposes a decision framework based on workload characteristics, isolation requirements, service targets, and operational capacity. Therefore, successful virtualization depends on architectural fit and disciplined management rather than platform branding alone.