Skip to content

README.md

Latest commit

 

History

History
82 lines (61 loc) · 3.97 KB

README.md

File metadata and controls

82 lines (61 loc) · 3.97 KB
Raw

NATS Client

A Python client for the NATS messaging system.

Features

  • Support for publish/subscribe
  • Support for request/reply
  • Support for queue groups
  • Support for multi-value message headers

Installation

pip install nats-client

Usage

import asyncio
from nats.client import connect

async def main():
    client = await connect("nats://localhost:4222")

    # Subscribe
    async with await client.subscribe("foo") as subscription:
        # Publish
        await client.publish("foo", "Hello World!")

        # Receive message
        message = await subscription.next()
        print(f"Received: {message.data}")

    await client.close()

if __name__ == "__main__":
    asyncio.run(main())

🚀 Performance

This client implementation delivers significant performance improvements over the nats.aio client, particularly for high-frequency, small message workloads.

Do note tho, it is not as feature complete at this point in time.

Message Size nats.py (python3) nats.py (pypy3) experimental-nats.py (python3) experimental-nats (pypy3) Performance Gain
1B 127,411 153,009 1,522,673 5,376,113 35.1x 🚀
2B 136,485 148,981 1,544,513 5,396,347 36.2x 🚀
4B 131,630 149,297 1,548,191 5,356,600 35.9x 🚀
8B 138,229 141,117 1,530,825 5,307,400 37.6x 🚀
16B 140,874 149,826 1,539,244 5,211,168 34.8x 🚀
32B 141,427 146,670 1,515,068 5,115,238 34.9x 🚀
64B 145,257 153,542 1,505,724 5,339,967 34.8x 🚀
128B 163,181 164,723 1,479,100 4,923,321 29.9x 🔥
256B 145,824 161,017 1,452,996 4,130,165 25.7x 🔥
512B 243,641 277,321 1,297,250 3,430,092 12.4x ⚡
1K 738,895 802,283 1,253,102 2,374,747 3.0x ⚡
2K 696,945 736,925 1,060,123 1,381,177 1.9x ✨
4K 577,335 625,935 798,797 814,393 1.3x ✨
8K 414,077 463,383 532,429 450,211 0.97x
16K 266,104 309,680 345,651 228,815 0.74x
32K 102,460 128,852 166,028 125,662 0.98x
64K 55,208 63,563 74,359 56,804 0.89x

Key Performance Insights

🎯 Sweet Spot: Small to Medium Messages

  • 35-37x faster for tiny messages (1B-64B)
  • 25-30x faster for small messages (128B-256B)
  • 12x faster for medium messages (512B)

Benchmark Environment

  • CPU: Apple M3 Max
  • Memory: 36 GB
  • Python: 3.x
  • PyPy: 3.x

Note: Benchmarks may vary based on your specific hardware, network conditions, and NATS server configuration. We recommend running your own benchmarks for production workloads.