HTTP/3 vs HTTP/2: A Comparative Analysis of Latency and Throughput in RESTful API Transactions Under Varying Concurrency, Payload, and Degraded Network Conditions
DOI:
https://doi.org/10.30871/jaic.v10i2.11803Keywords:
HTTP/3, HTTP/2, Network Performance, RESTful APIAbstract
This research empirically compares the performance of RESTful APIs based on HTTP/3 (quic-go) and HTTP/2 (Go standard library with TLS 1.3) under varying workload and network conditions. A rigorous quantitative factorial experiment design 2 × 2 × 2 × 2 × 3 was implemented in a controlled GCP Virtual Machine environment. The experiment tested two protocols, two REST methods, two payload sizes, two concurrency levels, and three network conditions (Baseline, Medium, and Poor). Network degradation was emulated using traffic control to simulate latency, bandwidth limitation, and packet loss. Performance was measured using Latency, Time to First Byte (TTFB), and Throughput. Non-parametric statistics (Mann-Whitney U Test, Kruskal-Wallis H Test, and ∈_rb Effect Size) were used for hypothesis testing. Results show that HTTP/3’s performance benefit is highly context-dependent. Under severely degraded network conditions, HTTP/3 exhibits a statistically and practically significant advantage, particularly for high-concurrency GET operations, achieving up to 31.2% lower latency and 45.4% higher throughput compared to HTTP/2. Across nearly all scenarios, HTTP/3 consistently delivers superior TTFB performance, averaging a 35.5% reduction, reflecting its efficient connection establishment and stream-level loss recovery. Conversely, under ideal network conditions with high concurrency and medium payload sizes, HTTP/2 outperforms HTTP/3, recording up to 29.3% lower latency and 22.7% higher throughput, suggesting implementation-level processing trade-offs in QUIC-based systems. These findings indicate that protocol selection should be guided by workload characteristics and network conditions rather than assumed protocol superiority. The results are implementation-dependent and derived from a controlled synthetic workload, which limits direct generalization to production environments.
Downloads
References
[1] J. Pendon and T. Warishe, “The Effects of HTTP Advances on Modern Page Loading,” Sci. Prepr., no. December, 2023, doi: 10.14293/PR2199.000570.v1.
[2] F. Liu, J. Dehart, J. Parwatikar, B. Farkiani, and P. Crowley, Performance Comparison of HTTP/3 and HTTP/2: Proxy vs. Non-Proxy Environments, vol. 1, no. 1. Association for Computing Machinery, 2024. [Online]. Available: http://arxiv.org/abs/2409.16267
[3] U. N. Hadianto, J. G. A. Ginting, D. Pranindito, and E. F. Cahyadi, “Throughput Assessment of HTTP/3 (QUIC)-Based Protocol on OpenLiteSpeed Web Server,” pp. 179–184, 2025, doi: 10.1109/ies67184.2025.11160983.
[4] J. Koch, O. Falowo, and N. Elrod, “What We Know About HTTP/3 and Its Implementation: A Literature Review,” 2024, doi: 10.1109/icmi60790.2024.10585883.
[5] M. Bishop, “Hypertext Transfer Protocol Version 3 (HTTP/3),” 2020, [Online]. Available: https://tools.ietf.org/html/draft-ietf-quic-http
[6] P. Tripathi, M. H. Miraz, and S. Joshi, “Comparing Communication Efficiency: HTTP/3 versus HTTP/1.1 Latency in RESTful APIs,” Proc. - 2023 Int. Conf. Comput. Networking, Telecommun. Eng. Sci. Appl. CoNTESA 2023, pp. 27–31, 2023, doi: 10.1109/CoNTESA61248.2023.10384951.
[7] M. Trevisan, D. Giordano, I. Drago, and A. S. Khatouni, “Measuring HTTP/3: Adoption and performance,” 2021 19th Mediterr. Commun. Comput. Netw. Conf. MedComNet 2021, pp. 1–8, 2021, doi: 10.1109/MEDCOMNET52149.2021.9501274.
[8] J. Dubec, J. Balazia, and P. Cicak, “Performance evaluation of the HTTP/3 client implementations,” 2023 46th Int. Conf. Telecommun. Signal Process. TSP 2023, pp. 260–263, 2023, doi: 10.1109/TSP59544.2023.10197834.
[9] X. Zak, J. Machaj, and L. Sevcik, “A Comparative Analysis of HTTP/2 and HTTP/3 Web Server Performance,” 15th Int. Conf. Elektro 2024, ELEKTRO 2024 - Proc., pp. 1–6, 2024, doi: 10.1109/ELEKTRO60337.2024.10556831.
[10] M. Seemann, “Adding a QUIC API for Go’s standard library TLS package,” 2023, libp2p Blog & News. [Online]. Available: https://blog.libp2p.io/2023-09-13-quic-crypto-tls/
[11] Thierry Beigbeder, “Developing and deploying an HTTP/3 API in Go,” 2025, [Online]. Available: https://blog.otvl.org/blog/http3-api-go-writing/
[12] T. Bressoud and D. White, “RESTful Application Programming Interfaces,” Springer, Cham, 2020, pp. 715–755. doi: 10.1007/978-3-030-54371-6_23.
[13] S. Kemp, “Digital 2024: 5 Billion Social Media Users,” We Are Soc., 2024, [Online]. Available: https://wearesocial.com/us/blog/2024/01/digital-2024-5-billion-social-media-users/
[14] V. Thatikonda, “Building Scalable and Performant Apis: Best Practices and Patterns,” J. Pharm. Res. Reports, pp. 1–3, Jan. 2022, doi: 10.47363/JPRSR/2022(3)146.
[15] P. Mesnard, “Data and Media Elements,” Apress eBooks, 2022, pp. 79–98. doi: 10.1007/978-1-4842-8983-9_4.
[16] D. Saif, C. H. Lung, and A. Matrawy, “An Early Benchmark of Quality of Experience between HTTP/2 and HTTP/3 using Lighthouse,” IEEE Int. Conf. Commun., 2021, doi: 10.1109/ICC42927.2021.9500258.
[17] D. Phanekham and S. Nair, “Accurate Bulk Throughput Benchmarks Using Reduced Network Resources,” IEEE/IFIP Netw. Oper. Manag. Symp., pp. 1–5, 2023, doi: 10.1109/NOMS56928.2023.10154370.
[18] G. Perna, M. Trevisan, D. Giordano, and I. Drago, “A first look at HTTP/3 adoption and performance,” Comput. Commun., vol. 187, pp. 115–124, 2022, doi: https://doi.org/10.1016/j.comcom.2022.02.005.
[19] Robert Bourgault, “Understanding the Power of Coefficient of Variation in Software Performance Testing,” 2023, [Online]. Available: https://dzone.com/articles/understanding-the-power-of-coefficient-of-variatio
[20] S. Sawilowsky, J. Sawilowsky, and R. Grissom, “Effect Size, The,” 2025, pp. 791–796. doi: 10.1007/978-3-662-69359-9_689.
[21] H. Dźwigoł and M. Dźwigoł–Barosz, “Exploring the methodological framework and importance of statistical analysis in quantitative management research,” Zesz. Nauk., vol. 2025, no. 222, pp. 211–225, 2025, doi: 10.29119/1641-3466.2025.222.12.
[22] M. Zhang, Z. Yan, K. Dong, and H. Li, “Research on HTTP/3: Performance, Characteristics, Issues, and Improvement Mechanisms,” pp. 399–404, 2025, doi: 10.1109/icetci64844.2025.11083994.
[23] R. Bonsu, K. O. Peprah, and W. Asiedu, “Analyzing the Impact of HTTP/3 on Web Application Performance,” J. Inf. Technol. Sci., vol. 10, no. 3, pp. 40–49, 2024, doi: 10.46610/joits.2024.v10i03.005.
[24] X. Zhang, A. Hassan, and Z. M. Mao, “QUIC is not Quick Enough over Fast Internet”, doi: 10.1145/3589334.3645323.
[25] P. Bi, LiteQUIC : Improving QoE of Video Streams by Reducing CPU Overhead of QUIC, vol. 1, no. 1. Association for Computing Machinery. doi: 10.1145/3664647.3681670.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Vidya Az Zahra, Rizka Ardiansyah
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License (Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) ) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
