600K concurrent HTTP connections with Clojure and http-kit

Add an endpoint with a poorly optimized SQL query: 20 concurrent connections.<p>I find these benchmarks to not be very valuable. Typically the bottlenecks are in the database, not in the web server/code. Even when it is in the code, optimizations only become truly valuable when you are at scale or doing something either horribly wrong or algorithmically unique.

Maybe I'm not understanding what is going on here but it appears the author's client is communicating over a socket on the same machine as the server. The author is seeing insanely high numbers because she/he is bypassing the entire TCP/IP stack.

Hey, http-kit's author here. willing to answer any questions.

Author here. The server is open source, on github: <a href="https://github.com/http-kit/http-kit" rel="nofollow">https://github.com/http-kit/http-kit</a> Test test code is also on github: <a href="https://github.com/http-kit/scale-clojure-web-app" rel="nofollow">https://github.com/http-kit/scale-clojure-web-app</a><p>Suggestions very welcome!

The server itself seems to be written in Java: <a href="https://github.com/http-kit/http-kit" rel="nofollow">https://github.com/http-kit/http-kit</a> I wonder what the factors were for not writing it in Clojure if Clojure is the target platform.

Similar article here, going to 11:<p><a href="http://www.metabrew.com/article/a-million-user-comet-application-with-mochiweb-part-1" rel="nofollow">http://www.metabrew.com/article/a-million-user-comet-applica...</a>

Feng (http-kit's author) should be around shortly if anyone has any questions.<p>In the meantime, you can also check out <a href="http://http-kit.org" rel="nofollow">http://http-kit.org</a> for more info. That page is a work-in-progress so please excuse any errors.<p>We were actually planning to post to HN later this week; seems someone beat us to the punch :-)

This reminds me of <a href="http://blog.whatsapp.com/index.php/2012/01/1-million-is-so-2011/" rel="nofollow">http://blog.whatsapp.com/index.php/2012/01/1-million-is-so-2...</a> where they handled 2 millions concurrent tcp connections.

Once You give a little bit more of real code into those requests your JVM will die from continous GC. It's simple math :)

Anybody heard about joxa? Think about erlang beam features on a clojure with scheme like simplicity.

I've created the same test app with Play 2.1 RC2 and Scala: <a href="https://github.com/jamesward/scale-play-web-app" rel="nofollow">https://github.com/jamesward/scale-play-web-app</a><p>On my laptop I have half as many cores as the poster and am getting about half the performance.

This looks really cool. But can Apache and other frameworks accomplish this?

What is memory usage per connection? How much unnecessary data copying is going on? What is latency?<p>What happens under the real load of, say, a thousand concurrent TCP connections <i>together</i> with few thousand of back-end/other data-sources pending calls?<p>What will happen to memory usage and latency when simple setup above will serve simplest <i>remote</i> requests (which means stalled connections, re-tranmitions, etc) for 8 hours? 24 hours?