.png)
4 months ago
19
Extremely simple comparisons between fast ASGI frameworks. Each framework used uvicorn as the server using uvicorn file:app --workers 4. Tests were performed on a Dell XPS 13 with 16GB of RAM.
from MicroPie import App
class Root(App):
async def index(self):
return {'hello': 'world'}
app = Root()
$ wrk -d15s -t4 -c64 http://127.0.0.1:8000
Running 15s test @ http://127.0.0.1:8000
4 threads and 64 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 1.35ms 1.09ms 21.59ms 82.45%
Req/Sec 13.24k 1.63k 17.55k 73.83%
791721 requests in 15.03s, 121.56MB read
Requests/sec: 52685.82
Transfer/sec: 8.09MB
from blacksheep import Application, get, json
app = Application()
@get("/")
def home():
return json({"hello": "world"})
$ wrk -d15s -t4 -c64 http://127.0.0.1:8000
Running 15s test @ http://127.0.0.1:8000
4 threads and 64 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 1.15ms 392.66us 15.11ms 85.91%
Req/Sec 13.88k 1.39k 26.76k 87.04%
831432 requests in 15.10s, 120.52MB read
Requests/sec: 55060.05
Transfer/sec: 7.98MB
from fastapi import FastAPI
app = FastAPI()
@app.get("/")
def read_root():
return {"Hello": "World"}
$ wrk -d15s -t4 -c64 http://127.0.0.1:8000
Running 15s test @ http://127.0.0.1:8000
4 threads and 64 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 3.52ms 1.82ms 56.73ms 91.17%
Req/Sec 4.71k 404.21 5.29k 95.00%
281493 requests in 15.01s, 38.12MB read
Requests/sec: 18756.73
Transfer/sec: 2.54MB
from sanic import Sanic
from sanic.response import json
app = Sanic("my-hello-world-app")
@app.route('/')
async def test(request):
return json({'hello': 'world'})
$ wrk -d15s -t4 -c64 http://127.0.0.1:8000
Running 15s test @ http://127.0.0.1:8000
4 threads and 64 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 1.84ms 0.97ms 37.83ms 92.93%
Req/Sec 8.95k 790.43 18.25k 96.68%
536203 requests in 15.10s, 82.33MB read
Requests/sec: 35508.58
Transfer/sec: 5.45MB
import muffin
app = muffin.Application()
@app.route('/')
async def example(request):
return muffin.ResponseJSON({'hello': 'world'})
$ wrk -d15s -t4 -c64 http://127.0.0.1:8000
Running 15s test @ http://127.0.0.1:8000
4 threads and 64 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 1.08ms 516.86us 29.21ms 91.45%
Req/Sec 14.88k 1.71k 42.57k 94.68%
889891 requests in 15.10s, 120.51MB read
Requests/sec: 58931.31
Transfer/sec: 7.98MB
from litestar import Litestar, get
@get("/")
async def hello_world() -> dict[str, str]:
"""Keeping the tradition alive with hello world."""
return {"hello": "world"}
app = Litestar(route_handlers=[hello_world])
$ wrk -d15s -t4 -c64 http://127.0.0.1:8000
Running 15s test @ http://127.0.0.1:8000
4 threads and 64 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 1.57ms 633.11us 33.66ms 88.42%
Req/Sec 10.20k 1.41k 32.86k 91.51%
610059 requests in 15.10s, 82.62MB read
Requests/sec: 40401.18
Transfer/sec: 5.47MB
from starlette.applications import Starlette
from starlette.responses import JSONResponse
from starlette.routing import Route
async def homepage(request):
return JSONResponse({'hello': 'world'})
app = Starlette(debug=True, routes=[
Route('/', homepage),
])
$ wrk -d15s -t4 -c64 http://127.0.0.1:8000
Running 15s test @ http://127.0.0.1:8000
4 threads and 64 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 1.22ms 387.05us 17.42ms 85.42%
Req/Sec 13.05k 1.05k 15.33k 84.83%
779092 requests in 15.00s, 105.51MB read
Requests/sec: 51930.45
Transfer/sec: 7.03MB
The table below summarizes the performance of various ASGI frameworks based on a 15-second wrk test with 4 threads and 64 connections, measuring a simple "hello world" JSON response.
| Muffin | 889,891 | 58,931.31 | 7.98 | 1.08 | 0.52 | 29.21 |
| Blacksheep | 831,432 | 55,060.05 | 7.98 | 1.15 | 0.39 | 15.11 |
| MicroPie | 791,721 | 52,685.82 | 8.09 | 1.35 | 1.09 | 21.59 |
| Starlette | 779,092 | 51,930.45 | 7.03 | 1.22 | 0.39 | 17.42 |
| Litestar | 610,059 | 40,401.18 | 5.47 | 1.57 | 0.63 | 33.66 |
| Sanic | 536,203 | 35,508.58 | 5.45 | 1.84 | 0.97 | 37.83 |
| FastAPI | 281,493 | 18,756.73 | 2.54 | 3.52 | 1.82 | 56.73 |
