Re:Up and Running with Clojure
by jarecare on Thursday October 20, 2011 @01:02PM
(#3135724)
I wonder if other JVM languages, e.g., Scala, would have the same performance in the "Hello World" test cited in the article.
[ Parent
| on HN ]
I'm the author of Aleph, the Clojure web server being benchmarked.
The most recent release of Aleph has focused on adding instrumentation and support for easily creating asynchronous workflows. A lot of new abstractions have been introduced, and "Hello World" performance has suffered a bit as a result.
However, now that all this has been introduced (0.2.0 is very close to release), I'm planning a "tick" release [1] that will focus on regaining some of the lost performance.
So please don't judge Clojure's potential for performant Hello World servers by what Aleph can do right now.
thank you for informing the numbers, I get much better results with straight ring and jetty through clojure, so it's annoying to see 'clojure is exactly this slow relative to the others'
(author here) I have the same question. We'd love someone to do it (esp. a Scala pro) and send us a pull request.
Re:Up and Running with Clojure
by swannodette on Thursday October 20, 2011 @02:42PM
(#3136232)
If you're going to benchmark, I recommend that you benchmark with the most common Clojure web setup - Ring + Jetty.
In fairness to the author, he only used event driven servers everywhere. The default assumption would probably be that this was the highest throughput approach, but ignores the sheer maturity of Jetty vs Aleph. I doubt the numbers for ruby would be as good if he'd used WEbrick.