Hosting
WordPress Web Server Benchmark: Reported September 2025 Results
This historical article preserves the measurements reported in QUAPE’s September 2025 comparison of NGINX, Apache and OpenLiteSpeed. It is not a new benchmark or a universal ranking of web servers. The original raw test files and complete configuration artefacts are not included here, so the numbers below should be read as reported results, not independently reproduced measurements.
What the original test described
The original article described three WordPress and WooCommerce environments, each with 16 CPU cores, 32,000 MB of RAM and 200 GB of SSD storage. It reported using Locust with 25 concurrent users, a ramp-up of five users per second and a ten-minute test duration.
Those details describe part of the test, but not enough to reproduce it. Software versions, web-server and cache configuration, the Locust workload, request mix, data set and load-generator placement are also important.
Reported measurements
| Metric | OpenLiteSpeed | NGINX | Apache |
|---|---|---|---|
| Total requests | 5,557 | 5,797 | 5,397 |
| Average response | 201.09 ms | 100.56 ms | 271.36 ms |
| Requests per second | 9.26 | 9.66 | 9.00 |
| Reported success rate | 99.98% | 100.00% | 100.00% |
| Maximum response | 4,049.71 ms | 1,091.94 ms | 4,224.07 ms |
| Minimum response | 1.33 ms | 2.03 ms | 1.55 ms |
| Failures | 1 | 0 | 0 |
What can be said about this sample
NGINX had the lowest reported average and maximum response times in this run. It did not have the lowest minimum response time. OpenLiteSpeed’s reported average was about 26% lower than Apache’s, not twice as fast as an earlier version of this article suggested.
Zero recorded failures during a ten-minute run is not proof of perfect reliability or future uptime. Likewise, a reported average server response is not a browser Core Web Vitals score or evidence of improved search rankings.
What a useful repeat test should include
- Exact software versions and configuration files for each environment.
- Equivalent caching conditions, including warm and cold runs where relevant.
- The test script, request mix and definition of success or failure.
- Several repeat runs and response-time percentiles, not only averages.
- Server resource measurements and load-generator capacity.
- Representative dynamic requests such as cart, account and checkout actions.
Keep the complete results and explain differences in setup before attributing an outcome to the web-server software alone. Do not apply one small historical sample to every WordPress workload.
Use workload evidence when choosing hosting
For QUAPE WordPress hosting, discuss the site’s actual functions, bottlenecks and support requirements. Our WooCommerce hosting checklist helps frame a store-specific evaluation. A suitable configuration and a tested maintenance process are more useful purchasing criteria than declaring one web server the winner for every site.
