[erlang-questions] Cowboy (Erlang) VS Haskell (Warp)

okeuday <>
Thu Jun 27 21:59:02 CEST 2013

To be a bit more accurate, when I hook-up elli to CloudI it shows that 
cowboy has better performance when doing the same http_req test with 
httperf.  So, it is likely that the benchmark (
https://github.com/yesodweb/benchmarks) could be faster with cowboy (i.e., 
CloudI is limiting throughput, probably mainly due to erlang:now/0 usage).  
One of the main reasons cowboy is faster is that the connection rate is an 
order of magnitude larger: 124.1 conn/s instead of 11.6 conn/s.  The 
throughput, for the curious was 12539.2 req/s cowboy and 11560.6 req/s 
elli.  I feel compelled to say again how httperf is not an accurate way to 
judge performance (see 
http://www.mnot.net/blog/2011/05/18/http_benchmark_rules), but it does 
provide a quick number to guess with.  There is more serious loadtesting of 
CloudI here 
https://github.com/okeuday/CloudI/tree/master/src/tests/http_req .

On Thursday, June 27, 2013 4:32:53 AM UTC-7, Loïc Hoguin wrote:
> On 06/27/2013 11:17 AM, Knut Nesheim wrote:
> > I happily help anybody who comes to me with questions.
> Me too, but I'm not sure how that would help someone trying to debug an 
> issue in production quickly to avoid losing money.
> You shouldn't make the product's usability dependent on the author's 
> availability. (And yep, I am also in the process of learning that, 
> except in my case it's with regards to documentation.)
> -- 
> Loïc Hoguin
> Erlang Cowboy
> Nine Nines
> http://ninenines.eu
> _______________________________________________
> erlang-questions mailing list
>  <javascript:>
> http://erlang.org/mailman/listinfo/erlang-questions
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20130627/36fae9d9/attachment.html>

More information about the erlang-questions mailing list