Since version 1.0, PanelAlpha Engine supports multiple web server configurations, giving hosting providers flexibility in how WordPress environments are deployed and optimized. Instead of forcing a single stack, the Engine allows operators to choose between different web servers depending on their performance, cost, and operational preferences.
To understand how these configurations behave in practice, we ran a comprehensive set of performance benchmarks across all supported web server variants on PanelAlpha Engine.
TL;DR The results are clear:
- LiteSpeed Enterprise and OpenLiteSpeed significantly outperform other configurations by a wide margin.
- OpenLiteSpeed delivers outstanding performance, in some metrics even surpassing the Enterprise version, which honestly surprised us.
- Caching plays a role, but not as much as you might expect. The benchmarks show that the web server itself is the dominant performance factor.
Test prerequisites and methodology
All benchmarks were executed using a consistent setup to ensure meaningful comparisons between configurations.
- The same WordPress installation and content were used across all platforms
- No CDN or external caching layers
- Default panel configurations were used for all platforms
- In the case of PanelAlpha Engine, the only variable changed was the web server
(LiteSpeed Enterprise, OpenLiteSpeed or the Nginx + Apache stack – which was the only option for BETA version) - Load testing tool: k6 with scenario:
- 20 concurrent virtual users
- ~20 seconds per test
- Requests distributed across the site using sitemap URLs
- 0 percent HTTP errors in all test runs
- All tests were performed on the same server plan on Hetzner: CPX23, Shared vCPU x86, 2vCPUS, 4 GB RAM, 40GB SSD
All raw benchmark outputs and detailed logs are publicly available here:
[link to raw results PDF]
The goal of these tests was to observe real behavior under concurrent load, not to tune individual configurations for benchmark optimization.
1. Average response time (ms)
Average response time represents how long it takes for a request to be fully processed and responded to. It is one of the most visible indicators of perceived performance.
Results:
| Web server configuration | Average response time (ms) |
|---|---|
| LiteSpeed Enterprise (non-cache) | ~54 |
| LiteSpeed Enterprise + Redis | ~63 |
| LiteSpeed Enterprise + LSCache | ~54 |
| LiteSpeed Enterprise + LSCache + Redis | ~55 |
| OpenLiteSpeed (non-cache) | ~54 |
| OpenLiteSpeed + Redis | ~53 |
| OpenLiteSpeed + LSCache | ~55 |
| OpenLiteSpeed + LSCache + Redis | ~53 |
| Nginx (non-cache) | ~288 |
| Nginx + Redis | ~273 |
| Apache (non-cache) | ~324 |
| Apache + Redis | ~308 |
| Nginx + Apache (default, non-cache) | ~365 |
| Nginx + Apache + Redis | ~340 |
LiteSpeed Enterprise and OpenLiteSpeed operate in a completely different performance class, delivering responses in tens of milliseconds. Nginx and Apache-based stacks are several times slower on the same Engine.
2. Average requests per second
Requests per second indicates how many concurrent requests the system can handle before becoming constrained.
Results:
| Web server configuration | Requests per second |
|---|---|
| LiteSpeed Enterprise (non-cache) | ~346 |
| LiteSpeed Enterprise + Redis | ~292 |
| LiteSpeed Enterprise + LSCache | ~325 |
| LiteSpeed Enterprise + LSCache + Redis | ~332 |
| OpenLiteSpeed (non-cache) | ~339 |
| OpenLiteSpeed + Redis | ~350 |
| OpenLiteSpeed + LSCache | ~334 |
| OpenLiteSpeed + LSCache + Redis | ~352 |
| Nginx (non-cache) | ~65 |
| Nginx + Redis | ~69 |
| Apache (non-cache) | ~58 |
| Apache + Redis | ~61 |
| Nginx + Apache (default, non-cache) | ~52 |
| Nginx + Apache + Redis | ~55 |
LiteSpeed-based configurations handle 5–7x more traffic than Nginx or Apache on the same Engine. Redis provides a modest improvement, but the web server choice is the primary driver of throughput.
3. Time to First Byte
TTFB measures how quickly the server starts responding after receiving a request.
Results:
| Web server configuration | Average TTFB (s) |
|---|---|
| LiteSpeed Enterprise (non-cache) | ~0.34 |
| LiteSpeed Enterprise + Redis | ~0.33 |
| LiteSpeed Enterprise + LSCache | ~0.23 |
| LiteSpeed Enterprise + LSCache + Redis | ~0.22 |
| OpenLiteSpeed (non-cache) | ~0.20 |
| OpenLiteSpeed + Redis | ~0.20 |
| OpenLiteSpeed + LSCache | ~0.20 |
| OpenLiteSpeed + LSCache + Redis | ~0.19 |
| Nginx (non-cache) | ~0.25 |
| Nginx + Redis | ~0.23 |
| Apache (non-cache) | ~0.25 |
| Apache + Redis | ~0.24 |
| Nginx + Apache (default) | ~0.26 |
TTFB remains low across all Engine configurations, but OpenLiteSpeed consistently deliver the fastest initial response times. Surprisingly, it’s LiteSpeed Enterprise didn’t
4. 90th percentile response time
The 90th percentile highlights how the system behaves under peak load conditions rather than average cases.
Results:
| Web server configuration | p90 response time (ms) |
|---|---|
| LiteSpeed Enterprise (non-cache) | ~58 |
| LiteSpeed Enterprise + Redis | ~63 |
| LiteSpeed Enterprise + LSCache | ~58 |
| LiteSpeed Enterprise + LSCache + Redis | ~63 |
| OpenLiteSpeed (non-cache) | ~59 |
| OpenLiteSpeed + Redis | ~58 |
| OpenLiteSpeed + LSCache | ~61 |
| OpenLiteSpeed + LSCache + Redis | ~56 |
| Nginx (non-cache) | ~329 |
| Nginx + Redis | ~308 |
| Apache (non-cache) | ~366 |
| Apache + Redis | ~354 |
| Nginx + Apache (default) | ~410 |
LiteSpeed-based stacks maintain low and predictable latency even under load. Nginx and Apache stacks exhibit significantly higher tail latency.
5. Average request duration
Average request duration reflects how long server resources are occupied while processing requests.
Results:
| Web server configuration | Avg request duration (ms) |
|---|---|
| LiteSpeed Enterprise (non-cache) | ~54 |
| LiteSpeed Enterprise + Redis | ~63 |
| LiteSpeed Enterprise + LSCache | ~54 |
| LiteSpeed Enterprise + LSCache + Redis | ~55 |
| OpenLiteSpeed (non-cache) | ~54 |
| OpenLiteSpeed + Redis | ~53 |
| OpenLiteSpeed + LSCache | ~55 |
| OpenLiteSpeed + LSCache + Redis | ~53 |
| Nginx (non-cache) | ~288 |
| Nginx + Redis | ~273 |
| Apache (non-cache) | ~324 |
| Apache + Redis | ~308 |
| Nginx + Apache (default) | ~365 |
LiteSpeed-based configurations free server resources significantly faster, enabling higher concurrency and better stability under sustained traffic.
Summary and recommendations
These benchmarks show that web server selection is the single most important performance decision when deploying PanelAlpha Engine.
LiteSpeed Enterprise and OpenLiteSpeed consistently deliver:
- The lowest latency
- The highest throughput
- The most predictable behavior under load
Redis and LSCache provide incremental improvements, but they do not change the overall performance class. The core difference comes from the web server itself.
Based on these results, we strongly recommend using LiteSpeed Enterprise, or at minimum OpenLiteSpeed, for production WordPress environments on PanelAlpha Engine.
Comments