Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Experimental Assessment of Containers Running on Top of Virtual Machines

Hossein Aqasizade, Ehsan Ataie, Mostafa Bastam

TL;DR

Evaluates performance trade-offs when OS-level virtualization (containers) runs atop hardware-level virtualization (VMs) across CPU, memory, disk, and network. The approach uses an experimental real-world setup with Xen PVHVM and KVM, testing LXC, Docker, and Podman containers, along with twelve container-on-VM configurations against native VMs and native containers; benchmarks include 7zip, STREAM, IOzone, Netperf, and MySQL OLTP. Key findings show native hardware generally outperforms all configurations, while Docker on KVM and LXC on Xen frequently deliver strong performance among container-on-VM setups; Xen PVHVM offers disk I/O advantages, and MySQL throughput favors Xen-based deployments. These results guide data-center architects in selecting configurations that balance isolation, performance, and resource efficiency, and point to future work on Kubernetes-based high availability and deeper runtime evaluations. Future work includes exploring high availability in Kubernetes clusters and broader runtime performance studies.

Abstract

Over the past two decades, the cloud computing paradigm has gradually attracted more popularity due to its efficient resource usage and simple service access model. Virtualization technology is the fundamental element of cloud computing that brings several benefits to cloud users and providers, such as workload isolation, energy efficiency, server consolidation, and cost reduction. This paper examines the combination of operating system-level virtualization (containers) and hardware-level virtualization (virtual machines). To this end, the performance of containers running on top of virtual machines is experimentally compared with standalone virtual machines and containers based on different hardware resources, including the processor, main memory, disk, and network in a real testbed by running the most commonly used benchmarks. Paravirtualization and full virtualization as well as type 1 and type 2 hypervisors are covered in this study. In addition, three prevalent containerization platforms are examined.

Experimental Assessment of Containers Running on Top of Virtual Machines

TL;DR

Evaluates performance trade-offs when OS-level virtualization (containers) runs atop hardware-level virtualization (VMs) across CPU, memory, disk, and network. The approach uses an experimental real-world setup with Xen PVHVM and KVM, testing LXC, Docker, and Podman containers, along with twelve container-on-VM configurations against native VMs and native containers; benchmarks include 7zip, STREAM, IOzone, Netperf, and MySQL OLTP. Key findings show native hardware generally outperforms all configurations, while Docker on KVM and LXC on Xen frequently deliver strong performance among container-on-VM setups; Xen PVHVM offers disk I/O advantages, and MySQL throughput favors Xen-based deployments. These results guide data-center architects in selecting configurations that balance isolation, performance, and resource efficiency, and point to future work on Kubernetes-based high availability and deeper runtime evaluations. Future work includes exploring high availability in Kubernetes clusters and broader runtime performance studies.

Abstract

Over the past two decades, the cloud computing paradigm has gradually attracted more popularity due to its efficient resource usage and simple service access model. Virtualization technology is the fundamental element of cloud computing that brings several benefits to cloud users and providers, such as workload isolation, energy efficiency, server consolidation, and cost reduction. This paper examines the combination of operating system-level virtualization (containers) and hardware-level virtualization (virtual machines). To this end, the performance of containers running on top of virtual machines is experimentally compared with standalone virtual machines and containers based on different hardware resources, including the processor, main memory, disk, and network in a real testbed by running the most commonly used benchmarks. Paravirtualization and full virtualization as well as type 1 and type 2 hypervisors are covered in this study. In addition, three prevalent containerization platforms are examined.
Paper Structure (17 sections, 9 figures, 3 tables)

This paper contains 17 sections, 9 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. 2.1. Virtual Machines
  4. 2.2. Containers
  5. 2.3. Containers and Virtual Machines
  6. Research Methodology
  7. 3.1. Software and Hardware Configuration
  8. 3.2. Benchmarks
  9. 3.3 Configuration Scenarios
  10. Results
  11. 4.1. Processor
  12. 4.2. Memory
  13. 4.3. Disk
  14. 4.4. Network
  15. 4.5. MySQL Database
  16. ...and 2 more sections

Figures (9)

  • Figure 1: General architecture of containers, virtual machines, and their combination
  • Figure 2: Compress-7zip test results
  • Figure 3: STREAM test results
  • Figure 4: IOzone test results: read operations
  • Figure 5: IOzone test results: write operations
  • ...and 4 more figures