You're right that Go/Java have async style similar to Erlang. But the majority of production systems today that these languages run is some sort of web application. In this area, the Erlang VM holds its own pretty well, especially for websockets [1]. In that study Elixir's memory usage is higher, but total connections were almost identical to Go.
It'd be great if there were better benchmarks for common use cases of various languages. Spring on the Java side tends to be heavy on reflection usage, which is orders of magnitude slower than JIT'ed JVM methods. Benchmarks like the benchmark game don't capture this. Still despite that the Erlang VM performs very well on the benchmarks game compared to other dynamic/scripting languages. Often it's easily 5-10 times faster than Python or Ruby. Given the parents comment, I'd argue many programmers who enjoy developing with dynamic languages can do so with Elixir with comparable performance to Go/Java for high concurrency web applications.
It's not near the bottom either which is the point. Actually Phoenix (the only Erlang base web framework I could find at the link) edges out quite a number of other frameworks, roughly ranking in the middle of the pack. Specifically Phoenix performed 32k req/s, in comparison to say the Go based Gin which does 51k req/s or a arguably more comparable setup of,Python3 Flask with full ORM at 13k. Spring only manages in 23k req/s. Nothing to write home about either way but clearly along the lines of my argument that Erlang/Beam can hold its own. Though some of the ruby frameworks are damn impressive seaming, huh.
However it gets more interesting if you look at the Latency tab. There Phoenix comes in with an average of 7.9 ms. In comparison Gin averages 5.8 ms, Spring at 12.1 ms, Flask Py3 at 23.1 or Py2 at 14.7 ms.
Where it's really interesting is looking at the max latency. Presuming this indicates roughly how 99th and 95th percentiles measurements would compare. In this category Phoenix comes in second with a max of 22.0 ms, behind only lib-mongodb at 20.8 ms. The lib-Mongolia is one of the fastest frameworks by raw req/s.
Appreciate the link to the benchmarks! Much more interesting, especially if you're concerned about max latency (and likely 99/95 th percentiles). In this case BEAM/Phoenix would let you plan capacity to minimize max latency fairly well as it appears very consistent.
But raw java servlets delivered 100+k req/s. It may mean that all this event loop/async/actors hype is overrated, and regular blocking approach can also deliver.
… and "Most (all?) large systems developed using Erlang make heavy use of C for low-level code, leaving Erlang to manage the parts which tend to be complex in other languages, like controlling systems spread across several machines and implementing complex protocol logic."
FAQ 1.4 What sort of problems is Erlang not particularly suitable for?
hi riku_iki, I just checked out the link akka.io and it says that it caters to java/scala only ... from your last statement I understood that it was meant for java and go.
Hey thanks for that link ... I was aware of go having coroutines available as part of the language but not that something similar to what erlang provided was available to the go system as well...
Erlang is much slower in raw performance than both go and jvm: https://benchmarksgame.alioth.debian.org/u64q/erlang.html
Async approache similar to Erlang was reproduced for Java and Go already: https://akka.io