l00ker
Concurrency in modern programming languages: Rust vs Go vs Java vs Node.js vs Deno
This popped up in my Google news feed last night…
Concurrency in modern programming languages: Rust vs Go vs Java vs Node.js vs Deno
That first thing that struck me about the article – given the title – is that Erlang and/or Elixir (I would think Elixir should considered a “modern programming language”) weren’t even mentioned, much less, even considered for a benchmark test.
Never the less, after reading through the article(s) (7 in all), I decided to tinker just for fun. Even though the benchmarks seem meaningless (as pointed out in one or more of the comments), I wondered how well Elixir would preform, so I created a simple Elixir app using only Cowboy w/Plug to serve the html file like the other examples in the benchmark tests.
mix new plug_test --sup
# lib/plug_test.ex
defmodule PlugTest do
import Plug.Conn
def init(options), do: options
def call(conn, _opts) do
Process.sleep(200)
conn
|> put_resp_header("connection", "keep-alive")
|> put_resp_content_type("text/html; charset=utf-8")
|> send_file(200, "hello.html")
end
end
# hello.html
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Hello!</title>
</head>
<body>
<h1>Hello!</h1>
<p>Hi from Elixir</p>
</body>
</html>
Notice the Process.sleep(200) in call/2 above. In the code the author wrote for rust, go, etc., he introduces a 2 second sleep every tenth request. I’m currently not aware of a way to do that (?), so I just decided to sleep each request process for 200 ms instead.
Here’s what I got after 2 back-to-back runs of ab -c 100 -n 10000 http://localhost:4000/
Server Software: Cowboy
Server Hostname: localhost
Server Port: 4000
Document Path: /
Document Length: 178 bytes
Concurrency Level: 100
Time taken for tests: 20.483 seconds
Complete requests: 10000
Failed requests: 0
Total transferred: 4090000 bytes
HTML transferred: 1780000 bytes
Requests per second: 488.20 [#/sec] (mean)
Time per request: 204.833 [ms] (mean)
Time per request: 2.048 [ms] (mean, across all concurrent requests)
Transfer rate: 194.99 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 0.5 0 5
Processing: 200 202 1.1 202 211
Waiting: 200 201 1.0 201 210
Total: 201 202 1.5 202 215
Percentage of the requests served within a certain time (ms)
50% 202
66% 202
75% 203
80% 203
90% 204
95% 205
98% 206
99% 209
100% 215 (longest request)
Server Software: Cowboy
Server Hostname: localhost
Server Port: 4000
Document Path: /
Document Length: 178 bytes
Concurrency Level: 100
Time taken for tests: 20.490 seconds
Complete requests: 10000
Failed requests: 0
Total transferred: 4090000 bytes
HTML transferred: 1780000 bytes
Requests per second: 488.04 [#/sec] (mean)
Time per request: 204.901 [ms] (mean)
Time per request: 2.049 [ms] (mean, across all concurrent requests)
Transfer rate: 194.93 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 0.4 0 4
Processing: 200 202 1.1 202 211
Waiting: 200 201 0.9 201 211
Total: 201 202 1.3 202 212
Percentage of the requests served within a certain time (ms)
50% 202
66% 202
75% 203
80% 203
90% 204
95% 204
98% 206
99% 208
100% 212 (longest request)
I ran the above benchmarks on my workstation PC with the following specs:
Ubuntu Linux 20.04.1
Intel(R) Core™ i7-4770K CPU @ 3.50GHz w/ 8 cores & 16GB of memory
Erlang: 24.2.1
Elixir: 1.13.2-otp-24
I know the article is almost pointless, but what are your thoughts?
Stuff like this seems to pop up all time as well as the flurry of articles with titles like The Top 10 Backend Programming Languages You Should Know In 2022 etc., that show up near the beginning of the year, most all of which, hardly ever mention Elixir. 
Most Liked
kartheek
If someone is trying to serve a static hello.html and benchmark it - they should also include nginx, apache in it 
.
Real fun is inside the comments section - don’t miss it.
All you demonstrated is that thread sleep works in all languages, and that overhead is significantly smaller than 200ms. You did not really benchmark the various language / server combos.
When you add 2 second delay every 10 requests you make the comparison totally meaningless. You are mesuring delays, not the code.
Reading the file in every loop mesures reading from disk, not actual program performance.
Author’s response - comment1, comment2.
Most of the performance and benchmark articles are pointless. I skim through these kind of articles and enjoy comments section.
l00ker
Yeah. That’s exactly what I do 99% of the time as well. It’s just that this one was about concurrency, and when I saw how it was being benchmarked, I was mostly curious to see how well the BEAM’s Process.sleep/1 stacked up against all the others! 
I don’t have an account there or I may have left a comment about Erlang/Elixir with a link to the video of Saša’s awesome talk The Soul of Erlang and Elixir
ityonemo
Last Post!
dimitarvp
HAHAHA 
That really made me laugh for solid half a minute.
I absolutely loved how defensive the author became in the comments: “I wanted to write something like this for a long time and it started dragging on so I just put something out the door”, which is a super weird statement! Not like he was a journalist on a deadline, right? I guess a lot of people are addicted to getting attention on the net, I don’t know.
And he says that on a topic where a ton of people would love more benchmarks, and were curious and were of course bound to rip it apart. Not sure what was he expecting. 
Popular in Discussions
Other popular topics
Latest on Elixir Forum
Sponsor Spotlight
Develop your skills with books, videos, and courses.
Categories:
Sub Categories:
Forums
Popular Tags
- #ecto
- #liveview
- #troubleshooting
- #learning-elixir
- #deployment
- #library
- #erlang
- #testing
- #genserver
- #mix
- #absinthe
- #remote-other
- #otp
- #plug
- #how-to-question
- #macros
- #postgres
- #channels
- #elixirconf
- #exunit
- #discussion
- #code-sync
- #javascript
- #podcasts
- #onsite
- #dialyzer
- #docker
- #authentication
- #umbrella
- #full-time-contract
- #podcasts-by-brainlid
- #ecto-query
- #elixir-ls
- #phoenix_html
- #iex
- #blog-post
- #graphql
- #genstage
- #ai
- #websockets
- #supervisor
- #elixirconf-us
- #advent-of-code
- #distillery
- #processes
- #forms
- #api
- #metaprogramming
- #security
- #hex
Our Sponsors
Build Elixir applications with speed and confidence.
Supporting innovation across the BEAM ecosystem.
We build reliable cloud platforms for business-critical systems.
Catch errors, track performance, monitor hosts and more.
Error tracking for Elixir devs who love to ship. Start your free account.
Practical resources that improve the lives of professional developers.
Develop your skills with books, videos, and courses.
The deployment platform built for Elixir: PaaS ease, VPS control.
The team behind Membrane, Popcorn, LiveDebugger. Available for hire.
Courses that'll move you from confusion to "Aha, now I get it!"