Download websites that use MariaDB

About MariaDB

A database is not supposed to be visible from outside. Nothing about a properly configured server should tell a passing crawler which engine answered the query behind the page, and that single fact ought to shape how you read every number here.

So start with the caveat rather than the count. As of September 2026 we detect MariaDB on 45,117 sites, and the figure is a floor rather than a measurement. Detection comes from what leaks: a hosting control panel that names its own stack, an error page nobody suppressed, a version banner some framework helpfully added, an installer artefact left in a public directory. Servers that are locked down properly are invisible to us by design, and locked down servers skew large. Anyone reading our count as market share will be wrong, and wrong in a predictable direction.

What survives the caveat is still worth having, because the leaks are not random. They cluster in shared hosting, in panel managed environments and in stacks that were installed by a script rather than assembled by a team. So the detectable population tells you less about engineering taste and more about hosting economics, which for most commercial questions is the more useful of the two.

The project's history explains why. MariaDB Foundation has maintained it under the GPL v2 since 2009, written in C++, as a community fork that kept close compatibility with the system it forked from. Compatibility was the whole point, and the consequence is that most installations were never chosen at all. A distribution shipped it, a panel provisioned it, a one click installer selected it. Presence therefore marks a default far more often than a preference, and defaults are exactly what you want to know when you are sizing a migration or a support market.

Geography follows the hosting rather than the audience. United States carries 41,373 of what we see and the .com zone leads the domain split, both of which track where volume hosting is sold and registered, not where the software is admired.

Who needs the list. Competing database vendors scoping displacement campaigns. Managed hosting companies looking for estates to absorb. Security teams that want to know which public facing stacks are leaking backend detail at all, since the leak itself is a finding independent of the engine behind it. Tooling vendors qualifying accounts that already run open infrastructure and will therefore buy open infrastructure tools.

One asymmetry is worth stating plainly. For most technologies we track, absence carries information. Here it carries almost none. A site that does not report a database engine is overwhelmingly likely to be running one anyway and simply not saying so, so the useful queries all live on the positive side of the data. Filter for what we found, cross reference it against the platform, the country and the rest of the row, and treat the missing majority as unknown rather than negative.