Web Framework Benchmarks

This article compares the performance of ~20 web frameworks. The tests are json serialization performance and querying a result from the database (one per request and multiple per request).

Even though all disclaimers about such tests apply, it is still interesting to read (and besides, if we don't compare frameworks this way, what have we left?).

Some interesting things I saw (again, taking this with a grain of salt that applies to all such generic comparisons):

  • Some of the Java frameworks rank higher on the multiple-queries-per-request test than other tests. Maybe the reason is that Java's better database integration pulls them up in such cases.
  • I was disappointed by node.js performance. While I expected JavaScript to be slower, the database access tests are supposed to target node's sweet spot where the main performance bottleneck is the database and for which node.js is supposed to be tuned.
  • Play!'s evented I/O probably doesn't work as well as advertised... (taking this
  • Ruby frameworks need some work...
  • If you want to use scripted languages, consider PHP, Clojure, Groovy.