Download websites that use OpenGSE

About OpenGSE

OpenGSE is Google's open implementation of a servlet container, released under Apache 2.0 and written in Java. The project came out of infrastructure built to run Google web applications, and was published as a lightweight way to serve Java servlets without the weight of a full application server. As a piece of software it never attracted the community that grew around Tomcat or Jetty, and the source has been quiet for a long time.

As of September 2026 we identify OpenGSE on 519,438 sites, a footprint far larger than the project's obscurity would suggest. The count comes from what a server reports about itself in its response headers rather than from anything a site owner chose to install, and a page delivered through Google infrastructure can announce this stack even though whoever published the page never picked a servlet engine and would not recognise the name.

Distribution supports that reading. 455,253 of the sites we see sit in United States, and .com leads the domain mix, a concentration that says more about where the serving infrastructure and its customers are registered than about a preference among Java developers in one country. Server software that engineers choose deliberately tends to spread more evenly across markets, because engineers everywhere read the same documentation.

For a researcher the practical value runs opposite to what the name suggests. Nobody licenses a technology dataset in order to study this servlet engine on its own merits. The entry is useful as a way of isolating a serving population, because the domains in it share a delivery path, and grouping them exposes which markets, which domain endings and which co-installed analytics or marketing tags travel with that path. Fingerprinting by response header is inferential in one important way, since a header can be rewritten by a proxy, stripped at a CDN, or left in place by a platform that has changed underneath, so the figure is a floor for sites that advertise the stack rather than a measurement of what is actually running.

A per-domain lookup answers whether one address returns the header, which is rarely the question anyone has. The downloadable list answers the population question instead: how many domains carry the signal, where they are registered, what they run alongside, and how the set moves between exports, which is how a migrating hosting platform can be told apart from a static one.