Laboratory Web Performance Benchmarking of React.js, SvelteKit, and Nuxt.js SPA/MPA Implementations on an Academic Conference Website
Abstract
Framework and architecture choices can determine whether academic conference systems remain discoverable, responsive, and accessible for bandwidth-diverse users. This study identifies the framework-architecture combination best suited for comparable deployments. It benchmarks laboratory Core Web Vitals-aligned metrics across six implementations of React.js, SvelteKit, and Nuxt.js, deployed as Single- and Multi-Page Application (SPA/MPA) variants reconstructed from the ICONBIT UNESA conference website. Because Interaction to Next Paint (INP) requires field data unavailable on these low-traffic subdomains, Total Blocking Time (TBT) was adopted as a laboratory proxy for interactivity. Lighthouse, PageSpeed Insights, and JMeter measurements were analyzed using Kruskal-Wallis and Dunn-Bonferroni tests, showing significant differences across all five metrics (p < 0.001). SvelteKit achieved the lowest Largest Contentful Paint (LCP) (mean 1.606–1.617 s) with zero TBT for its SSR-MPA variant, but SSR-based MPA deployments showed mobile cold-start penalties (mean LCP 17–18 s), while Nuxt SPA showed weak Slow 4G interactivity (TBT 554 ms). SvelteKit SSR-MPA is recommended only when this risk is mitigated through pre-warming or provisioned compute, with SvelteKit SPA as the safer static alternative otherwise. Findings are bounded by the single reconstructed website and Netlify free-tier environment evaluated.
Related
- On the convergence of dynamic implementations of Hamiltonian Monte Carlo and no U-turn samplers
- Risk-minimizing GUE-implementations of truthful reporting in a Dynamic Social Choice Problem
- Semidefinite Implementations of the Proximal Point Algorithm and Consequences for Progressive Decoupling in Optimization
- Residual Model Criticism for Fitted SDE Implementations
- A symmetric conference matrix of order 86
BibTeX
@article{fadhil2026laboratory,
title = {{Laboratory Web Performance Benchmarking of React.js, SvelteKit, and Nuxt.js SPA/MPA Implementations on an Academic Conference Website}},
author = {Muhammad Ariq Fadhil and Riska Dhenabayu and Hafid Kholidi Hadi and Thwin Htoo Aung},
journal = {Journal of Information Systems and Informatics},
year = {2026},
month = oct,
doi = {10.63158/journalisi.v8i5.1757},
url = {https://doi.org/10.63158/journalisi.v8i5.1757},
note = {CC BY},
}
RIS
TY - JOUR
TI - Laboratory Web Performance Benchmarking of React.js, SvelteKit, and Nuxt.js SPA/MPA Implementations on an Academic Conference Website
AU - Muhammad Ariq Fadhil
AU - Riska Dhenabayu
AU - Hafid Kholidi Hadi
AU - Thwin Htoo Aung
PY - 2026
DA - 2026/10/24
JO - Journal of Information Systems and Informatics
DO - 10.63158/journalisi.v8i5.1757
UR - https://doi.org/10.63158/journalisi.v8i5.1757
AB - Framework and architecture choices can determine whether academic conference systems remain discoverable, responsive, and accessible for bandwidth-diverse users. This study identifies the framework-architecture combination best suited for comparable deployments. It benchmarks laboratory Core Web Vitals-aligned metrics across six implementations of React.js, SvelteKit, and Nuxt.js, deployed as Single- and Multi-Page Application (SPA/MPA) variants reconstructed from the ICONBIT UNESA conference website. Because Interaction to Next Paint (INP) requires field data unavailable on these low-traffic subdomains, Total Blocking Time (TBT) was adopted as a laboratory proxy for interactivity. Lighthouse, PageSpeed Insights, and JMeter measurements were analyzed using Kruskal-Wallis and Dunn-Bonferroni tests, showing significant differences across all five metrics (p < 0.001). SvelteKit achieved the lowest Largest Contentful Paint (LCP) (mean 1.606–1.617 s) with zero TBT for its SSR-MPA variant, but SSR-based MPA deployments showed mobile cold-start penalties (mean LCP 17–18 s), while Nuxt SPA showed weak Slow 4G interactivity (TBT 554 ms). SvelteKit SSR-MPA is recommended only when this risk is mitigated through pre-warming or provisioned compute, with SvelteKit SPA as the safer static alternative otherwise. Findings are bounded by the single reconstructed website and Netlify free-tier environment evaluated.
SN - 2656-5935
SN - 2656-4882
ER -