web frameworks benchmark

This project was inspired by Mihai Cracan's article Web REST API Benchmark on a Real Life Application. I was curious how Golang compare to other languages in terms of performance.

Tested frameworks

.NET core
Golang mux
Java Spring
PHP Laravel
Node express
Python Django (with debug web server)
Python Django (with gunicorn, 12 workers)

App under test

Each app exposes the same 4 endpoints with different complexity:

Hello World - simply respond with {"hello": "world"}.
Computation - Compute first 10 000 Fibonacci numbers and respond with {"status": "done"}.
Simple Listing - List all records from countries table.
Complex Listing - List all users who visited France alongside with all visited countries. Users has many to many relationship with countries.

Benchmark results

Hello world requests per second

Computation requests per second

Simple listing requests per second

Complex listing requests per second

Hello world response time distribution

Computation response time distribution

Simple listing response time distribution

Complex listing response time distribution

Benchmark metodology

Benchmark was run on Google Cloud Platform n1-standard-4 instance with 4 vCPUs and 15GB RAM All apps and MySQL database was running simultaneously in Docker containers bootstrapped with docker-compose.yml Apache Benchmark was run on separate instance.

How to run benchmark?

This repo contains several bash scripts to run benchmark. Scripts was tested on Ununtu 18.04 LTS.

Clone this repository.
Run ./up.sh - This script will install docker, start mysql database, restore database dump from database.sql file, start all applications and run newman tests to ensure that everything is up and running.
Run ./ab.sh hostname - This script will run Apache Benchmark against hostname and save results to data directory
To generate charts you need to have gnuplot installed and run gnuplot timeseries.plt; gnuplot rps.plt charts will be saved to graphs directory.

yusefmaali/web-rest-api-benchmark