Flask, Tornado, Japronto performance test comparison

Do a simple test against the web framework Flask, Tornado, and Japronto

Test environment, virtual machine Ubuntu16.04 4 core CPU 8G memory

View server configuration

root@localhost:/home/frog/test# uname -a
Linux localhost 4.4.0-87-generic #110-Ubuntu SMP Tue Jul 18 12:55:35 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

root@localhost:/home/frog/test# cat /proc/cpuinfo | grep model\ name
model name	: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
model name	: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
model name	: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
model name	: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
root@localhost:/home/frog/test#

root@localhost:/home/frog/test# cat /proc/meminfo | grep MemTotal
MemTotal:        8175012 kB

The pressure test uses  wrk  software (the project address is  https://github.com/wg/wrk ), which is a pressure test software that can generate very high loads on a multi-core cpu machine and adopts a multi-threaded design.
We first use 10  threads and 10000  connections to test for  30  s

1，Flask

Server script: flask_test.py

from flask import Flask
app = Flask(__name__)

@app.route('/')
def hello_world():
    return 'Hello World!'

if __name__ == '__main__':
    app.run(host='0.0.0.0',port=8881)

Test Results

root@localhost:/home/frog# wrk -t10 -c10000 -d30s --latency "http://192.168.3.81:8881"
Running 30s test @ http://192.168.3.81:8881
  10 threads and 10000 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   353.26ms  285.36ms   1.91s    85.11%
    Req/Sec    60.07     72.79   565.00     90.04%
  Latency Distribution
     50%  224.33ms
     75%  597.30ms
     90%  666.42ms
     99%    1.69s 
  7360 requests in 30.10s, 1.17MB read
  Socket errors: connect 0, read 113, write 0, timeout 451
Requests/sec:    244.55
Transfer/sec:     39.64KB
root@localhost:/home/frog#

2，Tornado

Server script: tornado_test.py

import tornado.ioloop
import tornado.web

class MainHandler(tornado.web.RequestHandler):
    def get(self):
        self.write("Hello, world")

def make_app():
    return tornado.web.Application([
        (r"/", MainHandler),
    ])

if __name__ == "__main__":
    app = make_app()
    app.listen(8882)
    tornado.ioloop.IOLoop.current().start()

Test Results

root@localhost:/home/frog# wrk -t10 -c10000 -d30s --latency "http://192.168.3.81:8882"
Running 30s test @ http://192.168.3.81:8882
  10 threads and 10000 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.16s   436.92ms   2.00s    69.59%
    Req/Sec   131.43    174.27     1.11k    86.81%
  Latency Distribution
     50%    1.04s 
     75%    1.53s 
     90%    1.86s 
     99%    1.99s 
  24406 requests in 30.09s, 4.82MB read
  Socket errors: connect 0, read 0, write 0, timeout 7901
Requests/sec:    811.07
Transfer/sec:    163.96KB
root@localhost:/home/frog#

3 ， ready

Server script: japronto_test.py

from japronto import Application

def hello(request):
	return request.Response(text='Hello world')

app = Application()
app.router.add_route('/', hello)
app.run(debug=True,port=8808)

Test Results

root@localhost:/home/frog# wrk -t10 -c10000 -d30s --latency "http://192.168.3.81:8808"
Running 30s test @ http://192.168.3.81:8808
  10 threads and 10000 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   458.47ms  189.77ms   2.00s    64.91%
    Req/Sec     1.62k     1.46k    8.39k    72.49%
  Latency Distribution
     50%  438.24ms
     75%  589.58ms
     90%  708.52ms
     99%  889.29ms
  362614 requests in 30.10s, 31.47MB read
  Socket errors: connect 0, read 0, write 0, timeout 1572
Requests/sec:  12047.43
Transfer/sec:      1.05MB
root@localhost:/home/frog#

to sum up

From the results,

Comparative Results

frame	Number of requests per second (Requests/sec)
flask	244.55
tornado	811.07
already ready	12047.43

Obviously, Japronto requests per second performance is much higher than the other two.