Table of Contents
The idea behind this repository is to benchmark different languages implementation of HTTP server.
The application i tested is minimal: the HTTP version of the Hello World example.
This approach allows including languages i barely know, since it is pretty easy to find such implementation online.
If you're looking for more complex examples, you will have better luck with the TechEmpower benchmarks.
Please do take the following numbers with a grain of salt: it is not my intention to promote one language over another basing on micro-benchmarks.
Indeed you should never pick a language just basing on its presumed performance.
I have filtered the languages by single runtime (i.e. Java on JVM): this way i can focus on a specific stack, keeping it updated to the last available version/APIs. Where possible i just relied on the standard library, but when it is not production-ready (i.e. Ruby, Python) or where the language footprint is deliberately minimal (i.e. Rust).
Python 3.8.0 is installed via homebrew.
Python is a widely used high-level, general-purpose, interpreted, dynamic programming language.
Node.js version 13.0.1 is installed by official OSX package.
Dart version 2.6.1 is installed via homebrew.
Java JDK 13.0.0 is installed by official OSX package.
Java is a VM based, statically typed, general-purpose language that is thread safe, object-oriented and, from version 8, supports functional paradigms.
Crystal 0.31.1 is installed via homebrew.
Crystal has a syntax very close to Ruby, but brings some desirable features such as statically typing and ahead of time (AOT) compilation.
Nim 1.0.2 is installed via homebrew.
Rust language version 1.39.0 is installed by official package.
Rust is an AOT, garbage collector free programming language, preventing segfaults and granting thread safety.
I used wrk as the loading tool.
I measured each application server six times, picking the best lap (but for VM based languages demanding longer warm-up).
wrk -t 4 -c 100 -d30s --timeout 2000 http://0.0.0.0:9292
These benchmarks are recorded on a MacBook PRO 15 mid 2015 having these specs:
- macOS Mojave
- 2.2 GHz Intel Core i7 (4 cores)
- 16 GB 1600 MHz DDR3
RAM and CPU
I measured RAM and CPU consumption by using macOS Activity Monitor dashboard and recording max consumption peak.
For the languages relying on pre-forking parallelism i reported the average consumption by taking a snapshot during the stress period.
|Language||App Server||Requests/sec||RAM (MB)||CPU (%)|
|Elixir||Plug with Cowboy||45477.66||45.3||619.0|
|Ruby||Puma||56979.27||> 110||> 520|
|Python||Gunicorn with Meinheld||110583.42||> 40||> 380|
puma -w 8 -t 2 --preload servers/rack_server.ru
Gunicorn with Meinheld
cd servers gunicorn -w 4 -k meinheld.gmeinheld.MeinheldWorker -b :9292 wsgi_server:app
I used the cluster module included into Node's standard library.
I used the async HTTP server embedded into the Dart standard library and compiled it with
dart2native AOT compiler.
dart2native servers/dart_server.dart -k aot dartaotruntime servers/dart_server.aot
Plug with Cowboy
cd servers/plug_server MIX_ENV=prod mix compile MIX_ENV=prod mix run --no-halt
I tested Java by using Jetty with the non blocking IO (NIO) APIs.
cd servers/jetty_server javac -cp jetty-all-uber.jar HelloWorld.java java -server -cp .:jetty-all-uber.jar HelloWorld
I used Crystal HTTP server standard library, enabling parallelism by using the
crystal build -Dpreview_mt --release servers/crystal_server.cr ./crystal_server
To test Nim i opted for the httpbeast library: an asynchronous server relying on Nim HTTP standard library.
nim c -d:release --threads:on servers/httpbeast_server.nim ./servers/httpbeast_server
I used the HTTP ServeMux GO standard library.
go run servers/servemux_server.go
I tested Rust by using the Hyper, an HTTP implementation based on Tokio.io.
cd servers/hyper_server cargo run --release