3 ms·

This should just fall under the 'duh' category. Quite simply, Erlang is the gold standard for scalability (horizontal and vertical) and nearly everything else i

This should just fall under the 'duh' category. Quite simply, Erlang is the gold standard for scalability (horizontal and vertical) and nearly everything else is far #2. However, choosing a tech is much more than just raw numbers. There are so many factors at play and it shouldn't be taken lightly. That being said, we did something similar about 2 years ago and Erlang/Riak came out so far ahead of the competition for our needs and it wasn't even close. I did go back to my notes just recently and looked again with Node in mind. Though I didn't do the same bake off we did back then, I did a thorough exercise with Node to make an evaluation and, while a cool tech, would not have made the cut. Erlang/OTP w/ Riak is not hard to program (anyone can learn), takes so much of the guess work out of scalability and is easy to debug. Node isn't quite mature enough to handle this yet and coupled with the true lack of multi-core, can't quite hang with the big boys just yet. Erlang/OTP is clearly in a league of its own.

ozataman 15y ago
Gold standard for scalability? There are certainly others. Check out numbers for Haskell: http://snapframework.com/blog/2010/11/17/snap-0.3-benchmarks http://snapframework.com/blog/2010/11/17/snap-0.3-benchmarks http://www.yesodweb.com/blog/preliminary-warp-cross-language-benchmarks http://www.yesodweb.com/blog/preliminary-warp-cross-language... Would be interesting to repeat the author's experiment exactly, but in any case I would expect an order of magnitude upwards difference in performance. Haskell gives you non-blocking IO without the hassle of callbacks, multithreaded operation and more.
- justincormack 15y ago
  I wouldnt expect an order of magnitude better on the same benchmark actually, especially as the benchmark in question is for a single core. There is clearly some scope to improve over these, but I doubt it is a factor of 10. Run the benchmarks rather than making claims with no backup.

ozataman 15y ago

Gold standard for scalability? There are certainly others. Check out numbers for Haskell: http://snapframework.com/blog/2010/11/17/snap-0.3-benchmarks http://snapframework.com/blog/2010/11/17/snap-0.3-benchmarks http://www.yesodweb.com/blog/preliminary-warp-cross-language-benchmarks http://www.yesodweb.com/blog/preliminary-warp-cross-language... Would be interesting to repeat the author's experiment exactly, but in any case I would expect an order of magnitude upwards difference in performance. Haskell gives you non-blocking IO without the hassle of callbacks, multithreaded operation and more.

justincormack 15y ago
I wouldnt expect an order of magnitude better on the same benchmark actually, especially as the benchmark in question is for a single core. There is clearly some scope to improve over these, but I doubt it is a factor of 10. Run the benchmarks rather than making claims with no backup.