Without publishing the complete source for each test, these benchmarks are not very useful (the author does promise to publish sources for a new test in future). It appears that initially node.js was in 'development mode', so presumably the post initially showed Play beating node. With unreproducible benchmarks there are inevitably configuration errors made, or configuration tweaks that aren't applied.
Yeah. And it's Java and all but I think there is something pretty odd if the mean time per request for Play is 1.5 seconds.
Some truly bad graphs there, displaying response time on same axis as responses per second? Leaving in a bracket probably to contain outliers and flattening the rest of the graph?
How does the "Render - No IO" case work? Does it take a different code path? (E.g. render "No entries found" instead of iterating over each entry and rendering it?) Or, put differently, what are you measuring in the "Render - No IO" case?
The source is now on github. See the links at the top of the blog post.
> I’m unable to make the source code for the test apps available publicly at this time.
and
> I spent nearly equal amounts of time (under two hours) to build the same app on nodejs and Play.
One actually wonders: why?
That also caught my eye, I thought this is either some prolific 2 hour work or there is some sort of bias he doesn't want us to know about?
It is kind of odd, but at the beginning of the article he states that he originally wrote and did the tests as part of his work at Ebay.
Yes, correct - I'm not able to disclose the code. But there is more to come and I will be sharing the code on github.
The source is now on github. See the blog post for github links.
Those sure are some misleading bar charts.
What's misleading about them?
Please feel free to reproduce on your own. The source is on github. See the blog post for github links.