Hosting & Domain

LiteSpeed Was Not Faster: 260 Persian Sites Measured

Author: Reading time: 6 min 1 views

Map of this article

Jump straight to the part you came for.

Hosting & Domain8 sections
  1. 1A third of these sites will not tell you what they run

    Apache has effectively left this market: eight sites out of 260.

  2. 2Response times

    Across all 260 domains the median time to first byte was 0.53s, with the 25th percentile at 0.37s and the 75th at 0.93s.

  3. 3The honest part

    That table does not prove LiteSpeed is slow.

  4. 4What LiteSpeed actually gives you

    What you are buying is not the server's name.

  5. 5Something in the headers that had nothing to do with speed

    33 of the 260 sites announce their PHP version in an x-powered-by header.

  6. 6One number that came out well

    252 of 260 sites answered over HTTP/2.

  7. 7What this measurement cannot support

    Measured from Helsinki, not Tehran.

  8. 8Ask these instead of asking the server's name

    Ask which PHP versions are available and whether you can change them yourself.

Median time to first byte on sites announcing LiteSpeed: 0.63 seconds. On sites announcing nginx: 0.61 seconds. In this sample LiteSpeed was not faster.

We called 304 Persian domains that rank in Google's top ten and read their response headers. 260 answered. Here is what came back, and here is why the headline above is less useful than it looks.

LiteSpeed Was Not Faster: 260 Persian Sites Measured

A third of these sites will not tell you what they run

Server headerSitesShare
no header sent8533%
nginx4116%
LiteSpeed2811%
Cloudflare208%
ArvanCloud208%
BitNinja145%
Sotoon104%
Apache83%
IIS83%

Two things jump out. Apache has effectively left this market: eight sites out of 260. The «nginx or Apache» argument that Persian tutorials still stage is an argument from ten years ago.

And 85 sites send no Server header at all. That is deliberate and it is correct practice, because the header tells an attacker which vulnerability list to open. It also means you cannot learn what they run from the outside, which is worth remembering the next time someone shows you a survey of «what Iranian sites use».

Response times

Across all 260 domains the median time to first byte was 0.53s, with the 25th percentile at 0.37s and the 75th at 0.93s. 121 sites answered in under half a second, 59 took over one second, 27 took over two, and the slowest took 12.8.

Declared stackSitesMedian TTFB
Cloudflare200.28s
Sotoon100.48s
no header850.50s
ArvanCloud200.59s
nginx410.61s
BitNinja140.62s
LiteSpeed280.63s
IIS80.73s
Apache80.95s

The honest part

That table does not prove LiteSpeed is slow. It proves something more useful: from the outside you cannot separate server speed from site speed.

The number we captured is the sum of the network path from Helsinki, every redirect on the way, the TLS handshake, whether a cache was warm, the weight of the home page itself, the database work behind it, and finally the web server. The web server's share of that sum is the smallest part of it.

Which is exactly why a host writing «LiteSpeed hosting, up to X times faster» is making a claim nobody outside can disprove either. A claim that can be neither confirmed nor refuted is not a technical claim. It is a label.

The only thing these numbers say with confidence is that Apache was slowest in this sample, and that the gap between LiteSpeed and nginx is small enough to be moved by eight sites.

What LiteSpeed actually gives you

What you are buying is not the server's name. It is its cache module. Serving a prepared page before the request ever reaches PHP is a genuine difference, and against no cache at all the comparison is fair.

But that is not the comparison. The real one is server-level cache against a decent cache layer on nginx, and there the gap becomes very small. This site runs nginx with a file cache layer and our own numbers are fine.

Where the advantage is real: shared hosting, where you cannot install anything at server level and the bundled cache plugin does the work for you. Where it disappears: a virtual server, where you are choosing the cache anyway.

Something in the headers that had nothing to do with speed

33 of the 260 sites announce their PHP version in an x-powered-by header. The spread: 8.1 on eleven sites, 7.4 on ten, 8.3 on eight, and 8.4 and 8.2 on two each.

Ten sites currently ranking in Google tell anyone who sends a request that they are on PHP 7.4, whose security support ended years ago. That combination is being vulnerable and handing over the address at the same time.

If the header is on for your own site, switch it off. In nginx it is one line, fastcgi_hide_header X-Powered-By;, and in php.ini it is expose_php = Off. Nothing breaks. It is worth reading alongside the signs that a site is already in trouble.

One number that came out well

252 of 260 sites answered over HTTP/2. Only eight are still on HTTP/1.1. This is one of the few places where the Iranian market is not behind.

What this measurement cannot support

  • Measured from Helsinki, not Tehran. Harsh on servers inside Iran, generous to anything behind a global network.
  • One call per domain rather than three, so warm and cold caches are both in the numbers.
  • Home pages only, which are usually the heaviest and the most cached page a site has.
  • 44 domains did not answer at all.
  • The Server header can be set to anything. A site can print nginx and run something else.

Ask these instead of asking the server's name

Ask which PHP versions are available and whether you can change them yourself. Ask at what level caching sits and whether you can turn it off. Ask how many sites share a server, which nobody answers precisely, and watch how the question is handled. Ask how often backups run and where they are kept.

None of that is on a brochure and all of it affects your speed and your security more than the web server does. A more fundamental fork is the hosting operating system and what follows from it, and if you are early in this, where shared hosting genuinely runs out is a better starting question than the stack. For the control panel, we covered the practical differences between cPanel and DirectAdmin separately, and our own Linux hosting specification is written against the questions above rather than against a badge.

Last thing, said late as usual. If your home page weighs two megabytes, changing host rescues nothing. Servers are measured in milliseconds and pages in seconds. Building something light in the first place outperforms any server upgrade you can buy.

304 domains called on 19 August 2026 from a server in Helsinki, one GET and one HEAD each, browser user agent, redirects followed. 260 returned a 2xx or 3xx response.

Hossein Parto

IT engineer (Semnan University), technical SEO specialist, and PHP web architect with 12+ years of hands-on experience since 2014. Founder of RGB.ir and builder of four live web platforms.

Want us to put this knowledge to work for your business?

The RGB team professionally handles everything you just read about, for your own site. Start with a free consultation.

Comments & Questions

Have a question about this article? Ask, we'll answer.

No comments yet; be the first.

Write Your Comment

Your email won't be published. Comments are shown after review.