Synthetic micro-benchmark

The purpose to see the behaviour of different http-servers under high RPS. For this moment we compare different java based approaches + go:

• tomcat servlet + apache async http client
• netty approach
• helidon approach with green threads
• go

Within each of the projects put vegeta reports produced on my machine MacBook Pro M2

More info: https://medium.com/@yunussov/performance-spring-tomcat-mvc-vs-spring-netty-webflux-vs-go-vs-helidon-2dd9d3f0e433

Spring MVC

http://localhost:8081/performance-mvc/?delay=100

echo "GET http://localhost:8081/performance-mvc?delay=100" | vegeta attack -duration=10s -rate=1000 | vegeta report

Spring Webflux

http://localhost:8082/performance-webflux/?delay=100

echo "GET http://localhost:8082/performance-webflux?delay=100" | vegeta attack -duration=10s -rate=1000 | vegeta report

Go

http://localhost:8083/performance-go?delay=100

echo "GET http://localhost:8083/performance-go?delay=100" | vegeta attack -duration=10s -rate=1000 | vegeta report

Helidon SE

http://localhost:8084/performance-helidon?delay=100

echo "GET http://localhost:8084/performance-helidon?delay=100" | vegeta attack -duration=10s -rate=1000 | vegeta report

Load generator - vegeta

Scenario