Dynamic pages
Lower median time to first byte on LiteSpeed across the home page, a post and search, when every request was a genuine cache miss.
Ahosting Technical Operations test
Apache vs LiteSpeed for WordPress
LiteSpeed was roughly twice as fast for dynamic WordPress execution in our lab environments. Once both platforms served a correctly cached page, the result was effectively tied at about 11 ms.
Tested 23 August 2026. Re-run when a material stack change requires it.
24 Years
Trusted Since 2002
99.9% Uptime
Guaranteed SLA
30-Day
Money-Back Guarantee
24/7
Expert Helpdesk
Executive summary
Three findings, one caveat
We ran the same WordPress fixture on an Ahosting Apache lab and an Ahosting LiteSpeed lab. The result splits cleanly by whether the page had to be generated or could be served from cache.
Cached pages
Effectively tied. Once a correct full-page cache answered the request, the web server stopped being the variable worth arguing about.
Under concurrency
Throughput advantage on dynamic requests at 1 to 20 concurrent users, with p95 latency roughly half at every step.
Test environment
What the two labs actually were
Identical WordPress fixture on both: WordPress 7.1, PHP 8.2, Twenty Twenty-Five 1.5, 50 posts, 8 pages, 60 comments and 10 local images. Cloudflare was DNS-only, so no edge cache sat in front of either site.
- Web server
- Apache 2.4.67
- PHP path
- PHP-FPM
- Page cache
- WP Super Cache 3.1.1
- Database
- Local MariaDB 10.11
- Web server
- LiteSpeed Ent 6.3.6
- PHP path
- LSAPI with OPcache
- Page cache
- LiteSpeed Cache 7.9
- Database
- Remote MySQL 8.0.41
- CPU threads
- 48
- Memory
- 31.5 GB
- Protocol
- HTTP/1.1
- CPU threads
- 40
- Memory
- 64.0 GB
- Protocol
- HTTP/2
The hardware was not identical. Memory, database placement and protocol differ between the two nodes, so these numbers describe our environments rather than the two web servers in isolation. We say so here rather than in a footnote.
Dynamic results
When WordPress has to build the page
Every request used a unique URL to force a cache miss. 30 observations per route and platform, from three independent rounds. Lower is better.
Median dynamic TTFB
Lower is better
Apache
LiteSpeed
| Route | Apache median | LiteSpeed median | Reduction | Apache p95 | LiteSpeed p95 |
|---|---|---|---|---|---|
| Home page | 361.5 ms | 165.2 ms | 54.3% | 2,972.9 ms | 207.2 ms |
| Post | 376.5 ms | 175.7 ms | 53.3% | 2,583.4 ms | 217.0 ms |
| Search | 385.2 ms | 189.2 ms | 50.9% | 2,996.7 ms | 197.8 ms |
The p95 column is the more interesting one. Apache’s worst-case dynamic responses ran into seconds; LiteSpeed’s stayed close to its median. Consistency, not the headline average, is what a visitor actually feels.
Cached results
When the cache answers, the argument ends
Apache served from WP Super Cache in Expert mode via mod_rewrite. LiteSpeed served from server-integrated LSCache. Cache hits were verified rather than assumed.
Median TTFB on a cache hit
Lower is better
Apache
LiteSpeed
| Cacheable route | Apache median | LiteSpeed median | Apache p95 | LiteSpeed p95 |
|---|---|---|---|---|
| Home page | 11.1 ms | 10.9 ms | 14.4 ms | 11.7 ms |
| Post | 10.9 ms | 10.8 ms | 12.0 ms | 11.8 ms |
Concurrency
What happens when several people arrive at once
Both page caches were disabled for this run, so every request was dynamic. HTTPS with HTTP/1.1 keep-alive, from the same Michigan host.
Dynamic throughput
Higher is better
Apache
LiteSpeed
| Concurrency | Apache req/s | LiteSpeed req/s | Apache p95 | LiteSpeed p95 |
|---|---|---|---|---|
| 1 | 2.97 | 6.62 | 361 ms | 184 ms |
| 5 | 2.84 | 6.83 | 1,794 ms | 782 ms |
| 10 | 3.08 | 6.59 | 3,402 ms | 1,598 ms |
| 20 | 3.08 | 6.04 | 6,501 ms | 3,598 ms |
What it means
How to read this before you buy anything
The honest summary is narrower than the headline, and more useful.
Caching is the first decision
If your pages are cacheable and the cache is configured correctly, both stacks answer in about 11 ms. Get the cache right before you shop for a web server.
The gap is on cache misses
Logged-in sessions, search, personalised pages and anything a cache must skip. That is where the 51 to 54 percent showed up, and where a busy site actually lives.
Consistency beats the average
Apache p95 ran to seconds under load while LiteSpeed stayed near its median. The slowest requests are the ones a visitor remembers.
Limitations
What this study does not prove
Published so you can decide how much weight the numbers deserve.
- The two hosts did not use identical CPU counts, memory or database placement, so hardware is part of the result.
- Testing ran from a single Michigan location and does not represent global end-user latency.
- One WordPress fixture, one theme, no third-party plugin load. Real sites carry more.
- Results describe these Ahosting environments on the tested date, not a universal property of either web server.
Looking for store performance instead? The WooCommerce study measures cart, checkout and add-to-cart separately, and reaches a different conclusion.
Measured, then sold
The stack in this study is the one you get
Every WordPress plan runs on the LiteSpeed configuration measured above, not a different tier of it. WP Bronze starts at $2.79 a month and renews at $2.79, with a free dedicated IP, daily offsite backups, staging and a migration our team runs for you.
- LiteSpeed with LSCache on every plan
- Free dedicated IP included
- Free expert migration
- 30-day money-back period
WP Bronze starts from
$2.79 /month
36-month term. Renews at $2.79/mo, no increase and no setup fee.