Async Python Web Frameworks comparison

https://klen.github.io/py-frameworks-bench/

Updated: 2022-03-14

---

This is a simple benchmark for python async frameworks. Almost all of the frameworks are ASGI-compatible (aiohttp and tornado are exceptions on the moment).

The objective of the benchmark is not testing deployment (like uvicorn vs hypercorn and etc) or database (ORM, drivers) but instead test the frameworks itself. The benchmark checks request parsing (body, headers, formdata, queries), routing, responses.

Table of contents

• The Methodic
• The Results
  • Accept a request and return HTML response with a custom dynamic header
  • Parse path params, query string, JSON body and return a json response
  • Parse uploaded file, store it on disk and return a text response
  • Composite stats

The Methodic

The benchmark runs as a Github Action. According to the github documentation the hardware specification for the runs is:

• 2-core vCPU (Intel® Xeon® Platinum 8272CL (Cascade Lake), Intel® Xeon® 8171M 2.1GHz (Skylake))
• 7 GB of RAM memory
• 14 GB of SSD disk space
• OS Ubuntu 20.04

ASGI apps are running from docker using the gunicorn/uvicorn command:

gunicorn -k uvicorn.workers.UvicornWorker -b 0.0.0.0:8080 app:app

Applications' source code can be found here.

Results received with WRK utility using the params:

wrk -d15s -t4 -c64 [URL]

The benchmark has a three kind of tests:

1. "Simple" test: accept a request and return HTML response with custom dynamic header. The test simulates just a single HTML response.

2. "API" test: Check headers, parse path params, query string, JSON body and return a json response. The test simulates an JSON REST API.

3. "Upload" test: accept an uploaded file and store it on disk. The test simulates multipart formdata processing and work with files.

The Results (2022-03-14)

Accept a request and return HTML response with a custom dynamic header

The test simulates just a single HTML response.

Sorted by max req/s

Framework	Requests/sec	Latency 50% (ms)	Latency 75% (ms)	Latency Avg (ms)
blacksheep 1.2.5	18546	2.80	4.53	3.41
muffin 0.87.0	16571	3.09	5.17	3.83
sanic 21.12.1	15558	4.70	5.14	4.08
falcon 3.0.1	15554	3.29	5.49	4.08
baize 0.15.0	13880	3.69	6.21	4.58
starlette 0.17.1	13797	3.70	6.16	4.60
emmett 2.4.5	13380	5.54	6.10	4.75
fastapi 0.75.0	9060	5.46	9.79	7.03
aiohttp 3.8.1	7240	8.74	9.01	8.84
quart 0.16.3	3425	18.99	20.08	18.68
tornado 6.1	3232	19.76	19.94	19.81
django 4.0.3	1002	59.00	66.26	63.72

Parse path params, query string, JSON body and return a json response

The test simulates a simple JSON REST API endpoint.

Sorted by max req/s

Framework	Requests/sec	Latency 50% (ms)	Latency 75% (ms)	Latency Avg (ms)
sanic 21.12.1	10777	6.97	7.67	5.90
blacksheep 1.2.5	10505	4.70	8.16	6.07
muffin 0.87.0	10319	4.79	8.41	6.17
falcon 3.0.1	10133	4.88	8.61	6.28
starlette 0.17.1	8135	6.03	10.76	7.83
emmett 2.4.5	7091	7.17	11.58	9.12
baize 0.15.0	6581	9.96	10.24	9.71
fastapi 0.75.0	5882	8.36	15.16	10.85
aiohttp 3.8.1	4496	14.15	14.32	14.24
tornado 6.1	2780	22.95	23.17	23.02
quart 0.16.3	2146	29.42	30.05	29.81
django 4.0.3	883	68.00	71.74	72.37

Parse uploaded file, store it on disk and return a text response

The test simulates multipart formdata processing and work with files.

Sorted by max req/s

Framework	Requests/sec	Latency 50% (ms)	Latency 75% (ms)	Latency Avg (ms)
blacksheep 1.2.5	5604	8.87	15.77	11.40
sanic 21.12.1	5025	10.44	16.83	12.72
muffin 0.87.0	4425	11.14	19.99	14.43
falcon 3.0.1	3433	14.56	25.48	18.73
baize 0.15.0	2834	21.89	24.48	22.57
starlette 0.17.1	2434	20.10	36.39	26.26
aiohttp 3.8.1	2218	28.81	29.09	28.84
fastapi 0.75.0	2099	23.61	41.91	30.44
tornado 6.1	2067	30.89	31.09	30.95
quart 0.16.3	1746	36.68	37.58	36.63
emmett 2.4.5	1414	41.83	50.86	45.21
django 4.0.3	689	86.45	89.44	92.51

Composite stats

Combined benchmarks results

Sorted by completed requests

Framework	Requests completed	Avg Latency 50% (ms)	Avg Latency 75% (ms)	Avg Latency (ms)
blacksheep 1.2.5	519825	5.46	9.49	6.96
sanic 21.12.1	470400	7.37	9.88	7.57
muffin 0.87.0	469725	6.34	11.19	8.14
falcon 3.0.1	436800	7.58	13.19	9.7
starlette 0.17.1	365490	9.94	17.77	12.9
baize 0.15.0	349425	11.85	13.64	12.29
emmett 2.4.5	328275	18.18	22.85	19.69
fastapi 0.75.0	255615	12.48	22.29	16.11
aiohttp 3.8.1	209310	17.23	17.47	17.31
tornado 6.1	121185	24.53	24.73	24.59
quart 0.16.3	109755	28.36	29.24	28.37
django 4.0.3	38610	71.15	75.81	76.2

Conclusion

Nothing here, just some measures for you.

License

Licensed under a MIT license (See LICENSE file)