Performance Benchmarks

Independent, verifiable performance testing using industry-standard tools. All benchmarks are reproducible and conducted with production configurations.

uWebSockets.js

Maximum performance with uWebSockets.js transport layer.

Requests/sec226,253

Avg Latency3.92ms

Throughput32.1 MB/s

Peak Performance

• uWebSockets.js transport

• Same Moro API

• Lowest latency

Clustering Mode

Built-in clustering with auto worker management.

Requests/sec190,717

Avg Latency4.89ms

Throughput35.7 MB/s

Multi-Core

• Multi-core utilization

• Auto worker scaling

• Load distribution

Single-Threaded

Single process performance with production-ready configuration.

Requests/sec93,992

Avg Latency10.14ms

Throughput17.6 MB/s

Single Core

• Single process mode

• Minimal overhead

• Easy deployment

Framework Comparison

Every row below is sourced — pulled straight from the framework's own published benchmark, all on the same autocannon -c100 -d40 -p10 profile (Elysia run on Bun, NestJS via Sharkbench).

Framework	Req/sec	Avg Latency	Runtime	Source
NestJS (Fastify adapter)	9,164	3.5ms	Node 22	Sharkbench 2025
Express 5.2.1	28,104	35.07ms	Node 24	fastify/benchmarks
Koa 3.2.0	36,152	27.15ms	Node 24	fastify/benchmarks
Hono 4.12.9	37,320	26.30ms	Node 24	fastify/benchmarks
Fastify 5.8.4	46,193	21.14ms	Node 24	fastify/benchmarks
Elysia 1.4	71,202	12.6ms	Bun 1.1	APIScout 2026
MoroJS — single core	93,992	10.14ms	Node 22	Moro-JS/benchmark
MoroJS — clustered	190,717	4.89ms	Node 22	Moro-JS/benchmark
MoroJS — uWebSockets.js	226,253	3.92ms	Node 22	Moro-JS/benchmark

vs Fastify · uWS+390%226k vs 46k

vs Elysia (Bun)+218%226k vs 71k

vs Hono · clustered+411%190k vs 37k

vs Express · clustered+579%190k vs 28k

All framework numbers cite their original source. MoroJS variants from the published suite (M2 Ultra, Node 22.12). Click any source link to verify.

Key Findings

uWebSockets.js vs Clustered

Throughput Gain+19% faster

Latency Reduction~20% lower

Clustering vs Single-Threaded

Throughput Gain+103% (2x)

Latency Reduction~52% lower

Latency Stability~87% tighter

Detailed Performance Metrics

uWebSockets.js Statistics

Latency (ms)

Requests/Sec

2.5%: 1ms

1%: 215,935

50%: 4ms

50%: 226,559

97.5%: 6ms

97.5%: 231,807

Avg: 3.92ms

Avg: 226,252.8

Clustering Mode Statistics

Latency (ms)

Requests/Sec

2.5%: 4ms

1%: 174,079

50%: 5ms

50%: 191,743

97.5%: 5ms

97.5%: 193,535

Avg: 4.89ms

Avg: 190,716.8

Single-Threaded Statistics

Latency (ms)

Requests/Sec

2.5%: 6ms

1%: 86,655

50%: 8ms

50%: 93,951

97.5%: 17ms

97.5%: 97,279

Avg: 10.14ms

Avg: 93,992

Testing Methodology

Tools & Environment

Autocannon (industry standard)
MacBook Studio M2 Ultra
Node.js v22.12.0
Production configuration

Test Parameters

100 concurrent connections
40 second duration
10 pipelined requests
Multiple test runs

Test Configurations

uWebSockets.js (port 3112)
Clustered mode (port 3111)
Single-threaded (port 3110)
Same autocannon profile

Important Note

These are synthetic benchmarks designed to measure framework overhead. Real-world performance depends on your application logic, database queries, external API calls, and other factors. Always benchmark your specific use case.

Reproduce These Results

Clone & Setup

git clone https://github.com/Moro-JS/benchmark.git

cd benchmark && npm install

uWebSockets.js (Peak Performance) — port 3112

# Terminal 1 — start the uWS server

npm run hello-world-uws

# Terminal 2 — run the benchmark

npm run bench-uws

Clustered Mode — port 3111

# Terminal 1 — start the clustered server

npm run hello-world

# Terminal 2 — run the benchmark

npm run bench

Single-Threaded — port 3110

# Terminal 1 — start the single-threaded server

npm run hello-world-single

# Terminal 2 — run the benchmark

npm run bench-single

All Available Scripts

Server:npm run hello-world

Bench:npm run bench

Server:npm run hello-world-single

Bench:npm run bench-single

Server:npm run hello-world-uws

Bench:npm run bench-uws

Quick runs (shorter duration):

npm run bench-quick / bench-single-quick / bench-uws-quick