Actix(Rust) VS Netty(Java) VS Fastify(Nodejs)

actix_netty_vs_fastify.md

Fastify did 6K RPS (Request per Second)
Netty did 23K RPS (Request Per Second)
Actix did 53K RPS (Request Per Second)

Actix 1ms latency
Netty 4ms latency
Fastify 14ms latency

npm install -g autocannon

Actix

git clone https://github.com/diegopacheco/rust-playground.git
cd rust-microservice/
cargo build --release
./target/release/news-service

autocannon -c 100 -d 40 -p 10 172.17.0.2:8080/

Netty

docker run diegopacheco/time-microservice

autocannon -c 100 -d 40 -p 10 172.17.0.2:8080/rest/datetime

Fastify

git clone https://github.com/diegopacheco/Diego-Pacheco-Sandbox.git
cd scripts/node.js/fastify-fun2/
npm install
npm start

autocannon -c 100 -d 40 -p 10 localhost:3000

Fastify can certainly do better. 56254 req/sec with their benchmarks : https://www.fastify.io/benchmarks/

I don't know what you've been testing on your side. On my local machine (Apple M1 Pro) I did the following benchmarks:

Rust webserver (190k RPS):

Running the Rust implementation with .workers(10) (same as # of workers in Node) did not improve the RPS.

Is there anything else you could do to improve the RPS that I missed?

---

Single thread Fastify Webserver with Node v16.17 (146k RPS):

---

Fastify Webserver in Cluster Mode with Node v18.13 (191k RPS):