CoChat

A Comparative Analysis of OpenVPN, WireGuard, and IKEv2/IPSec Using Ubuntu 24.04 LTS

Open access
Ahmad Tamim Ahmadi, Sebghatullah Aslamzai · 2026 · Journal of Advanced Computer Knowledge and Algorithms

The DOI matches its Crossref record; 23 of the 26 references it cites were found in a registry or an index; none of the 56 sources its claims cite is checked for support yet.

Abstract

This paper presents a rigorous comparative analysis of three widely-used VPN protocols — OpenVPN, WireGuard, and IKEv2/IPSec — implemented and benchmarked on Ubuntu 24.04 LTS within a controlled VirtualBox virtualised environment on a Dell Inspiron 15 with 8 GB RAM. Each protocol was evaluated across throughput (iperf3, 30-second runs, 10 iterations), latency (ping, 100 packets), CPU utilisation, and packet jitter. Security was assessed via a structured feature matrix covering encryption, forward secrecy, authentication flexibility, formal verification, code complexity, and DPI resistance. WireGuard achieves the highest throughput (22.9 Mbps mean), lowest latency (4.7 ms), and minimal CPU footprint (9.8% server, 8.3% client). IKEv2/IPSec PSK delivers strong mid-range performance (16.8 Mbps, 8.2 ms). OpenVPN TCP/443 provides unique deep-packet-inspection evasion at a performance cost (8.7 Mbps, 14.3 ms). All three protocols provide AES-256-class encryption and perfect forward secrecy. Deployment scenario recommendations are provided for enterprise, performance-critical, and censorship-circumvention contexts.

Overview

CoChat Summary

TL;DR

This paper benchmarks OpenVPN, WireGuard, and IKEv2/IPSec on Ubuntu 24.04 LTS in a controlled VirtualBox environment, measuring throughput, latency, CPU utilisation, jitter, and security features. WireGuard showed the best performance, IKEv2/IPSec was mid-range, and OpenVPN TCP/443 was slowest but uniquely DPI-resistant.

Key findings

  • WireGuard achieved the highest mean throughput (22.9 Mbps), lowest mean latency (4.7 ms), lowest CPU utilisation (9.8% server / 8.3% client), and lowest jitter (0.8 ms mean, 1.6 ms max).
  • IKEv2/IPSec PSK delivered 16.8 Mbps and 8.2 ms, while IKEv2/IPSec Certificate delivered 14.3 Mbps and 9.1 ms; CPU was 18.6%/16.2% and 20.1%/18.4% for server/client respectively.
  • OpenVPN TCP/443 delivered 8.7 Mbps and 14.3 ms with 31.2%/28.9% CPU and 3.8 ms mean jitter; OpenVPN UDP/1194 delivered 11.2 Mbps and 11.8 ms with 28.4%/26.1% CPU and 2.9 ms mean jitter.
  • All three protocols provide AES-256-class encryption and perfect forward secrecy; WireGuard uniquely has formal cryptographic verification, IKEv2 offers EAP authentication flexibility, and OpenVPN TCP/443 uniquely provides DPI resistance.

Methods

  • Quantitative experimental design using two Ubuntu 24.04 LTS virtual machines in a VirtualBox host-only network (192.168.56.0/24) on a Dell Inspiron 15 with 8 GB RAM and kernel 6.8.x.
  • Configurations: WireGuard with Noise_IK/Curve25519 on UDP 51820; OpenVPN UDP/1194 and TCP/443 with EasyRSA 3.x, AES-256-GCM, TLS 1.3, HMAC-SHA256; strongSwan IKEv2/IPSec in PSK and X.509 certificate modes; OpenVPN compression disabled.
  • Performance measured with iperf3 (30-second TCP tests, 10 iterations), ping (100 ICMP packets, 10 runs), top/vmstat (1-second intervals), and iperf3 UDP jitter (1 Mbps target, 10 runs); security assessed via a six-dimension feature matrix scored 1–10 from published literature.

Limitations

  • The paper states future work should benchmark on physical hardware to eliminate hypervisor overhead, evaluate ARM, test WireGuard mesh configurations, explore post-quantum extensions, and conduct long-duration stability testing (24+ hours).
  • Evident from the text: results come from a virtualised environment with limited VM resources, so absolute throughput values are much lower than prior bare-metal benchmarks (e.g., 22.9 Mbps here vs ~1011 Mbps in Donenfeld), and may not generalise to other hardware or real-world networks.
  • Evident from the text: the security evaluation is a literature-based feature matrix scored 1–10, not empirical attack testing.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.29103/jacka.v3i3.27573

Cites

23 of 26 references verified0 citation errors · 3 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 VPN Tunneling Architecture Client-Server Communication Model.
Figure 1 VPN Tunneling Architecture Client-Server Communication Model.CC-BY-SA · From the PDF
Figure 2 Protocol Throughput Comparison (iperf3, 30-second runs, 10 iterations each).
Figure 2 Protocol Throughput Comparison (iperf3, 30-second runs, 10 iterations each).CC-BY-SA · From the PDF
Figure 3 Protocol Latency Comparison (ping average ± std dev, 100 packets).
Figure 3 Protocol Latency Comparison (ping average ± std dev, 100 packets).CC-BY-SA · From the PDF
Fig. 4 CPU Utilization During Data Transfer Server vs. Client.
Fig. 4 CPU Utilization During Data Transfer Server vs. Client.CC-BY-SA · From the PDF
Fig. 5 Security Feature Comparison Radar Chart (Score out of 10).
Fig. 5 Security Feature Comparison Radar Chart (Score out of 10).CC-BY-SA · From the PDF

Verification

Claims the paper makes

0 Valid · 0 Not valid · 37 Not checked · 16 Own results and interpretations

28 more claims

Create a free account to see all 53 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 14
  • Probably not a substantial claim. 10

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

2 of 2

Citations

Cites

Show all 26 references

Research works printed but not linked: 8

  1. F. Hauser et al., "A performance comparison of WireGuard and OpenVPN," in Proc. IEEE/IFIP NOMS, 2020.
  2. R. A. Ramadan, M. A. Azer, A. Farouk, and D. Hamza, "VPN security: Issues and challenges in developing countries," J. Inf. Secur. Appl., vol. 61, p. 102900, 2021.
  3. H. Nikkhah and V. Grover, "Companies' responses to data breaches: A longitudinal study," MIS Q., vol. 46, pp. 2163-2195, 2022. https://doi.org/10.25300/MISQ/2022/16220 · doi:10.25300/misq/2022/16220
  4. S. D. Warren and L. D. Brandeis, "The right to privacy," Harvard Law Rev., vol. 4, no. 5, pp. 193-220, 1890.
  5. R. A. Razaghpanah, R. Nithyanand, N. Vallina- Rodriguez, et al., "Apps, trackers, privacy, and regulators," in Proc. NDSS Symp., 2018.
  6. D. Kleine, ICTs and the Capability Approach. MIT Press, 2013.
  7. A. Martin, R. Lara-Cabrera, and D. Camacho, "Android malware detection through hybrid features fusion and ensemble classifiers," Inf. Fusion, vol. 52, pp. 128-142, 2019.
  8. J. Malik, A. Akhunzada, I. Bibi, et al., "Towards forensics-as-a-service for real-time remote evidence acquisition in software-defined networking," IEEE Access, vol. 8, pp. 134695-134706, 2020.

Cited by

Still being checked.

Metadata

Authors
Ahmadi, A. T., & Aslamzai, S.
Affiliations
Kabul University (AF)
Corresponding author
Ahmad Tamim Ahmadi
Venue
Journal of Advanced Computer Knowledge and Algorithms
Published
2026-06-20
Volume
3
Issue
3
Pages
96-102
ISSN
3031-8955 (ISSN-L)
OpenAlex
W7168386251
Cited by
0
References
22
FWCI
0.00
Type
journal article
Language
en
Open access
Locations
Journal of Advanced Computer Knowledge and Algorithms (publishedVersion, cc-by-sa)
Topics
  • Physical Sciences › Computer Science › Computer Networks and Communications › Cybersecurity and Information Systems
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › IPv6, Mobility, Handover, Networks, Security
  • Physical Sciences › Computer Science › Computer Networks and Communications › Network Security and Intrusion Detection
Keywords
OpenVPN, WireGuard, VPN protocols, throughput, latency, CPU utilization, packet jitter, forward secrecy, encryption, censorship circumvention, enterprise deployment

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-21, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-21 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.