TinyQ

A blazing fast, in-memory message queue in ~400 lines of C.

Features

• ⚡ Fast — 100,000+ ops/sec
• 🪶 Tiny — ~400 lines of C, ~5MB Docker image
• 🔒 Thread-safe — per-queue locking for concurrent access
• 🔌 Simple protocol — text-based TCP, easy to implement clients
• 📦 No dependencies — just libc and pthreads

Quick Start

# Build & run
make && ./tinyq-server

# Or with Docker
docker build -t tinyq .
docker run -p 7878:7878 tinyq

JavaScript Client

const { TinyQ } = require('./tinyq.js');

const q = new TinyQ({ host: 'localhost', port: 7878 });

await q.enqueue('jobs', 'Process payment #1234');
await q.enqueue('jobs', 'Send confirmation email');

const messages = await q.list('jobs'); // ['Process payment #1234', 'Send confirmation email']
const first = await q.dequeue('jobs'); // 'Process payment #1234'

// Listen for messages (polling)
const stopListening = q.listen({
    queue: 'jobs',
    concurrency: 10,
    onMessage: async (msg) => {
        console.log('Received:', msg);
        await handle(msg);
        console.log('Done:', msg);
    },
});

stopListening();

Protocol

Dead simple text protocol over TCP. Easy to implement in any language.

ENQUEUE <queue>\n<length>\n<payload>    →  OK\n
DEQUEUE <queue>\n                       →  OK <length>\n<payload>
LIST <queue> [limit]\n                  →  OK <count>\n[<len>\n<payload>]...

Benchmarks

Tested on MacBook Pro 2020 M1 (8-core, 16GB RAM):

Clients	Operations	Time	Throughput
10	20,000	0.24s	~84,000 ops/sec
25	50,000	0.50s	~99,000 ops/sec
50	100,000	0.87s	~115,000 ops/sec
100	200,000	1.70s	~117,000 ops/sec

Run your own benchmark:

CLIENTS=50 MESSAGES=1000 node bench.js

License

MIT