Is that benchmark reliable - aiohttp vs requests

问题

We are trying to chose between technologies at my work. And I thought I'd run a benchmark using both libraries (aiohttp and requests).

I want it to be as fair / unbiased as possible, and would love a look from the community into this.

So this is my current code :

import asyncio as aio
import aiohttp
import requests
import time

TEST_URL = "https://a-domain-i-can-use.tld"

def requests_fetch_url(url):
    with requests.Session() as session:
        with session.get(url) as resp:
            html = resp.text

async def aio_fetch_url(url):
    async with aiohttp.ClientSession() as session:
        async with session.get(url) as resp:
            html = await resp.text()

t_start_1 = time.time()
for i in range(10):
    [requests_fetch_url(TEST_URL) for i in range(16)]
t_end_1 = time.time()
print("using requests : %.2fs" % (t_end_1-t_start_1))

t_start_2 = time.time()
for i in range(10):
    aio.get_event_loop().run_until_complete(aio.gather(
        *[aio_fetch_url(TEST_URL) for i in range(16)]
    ))
t_end_2 = time.time()
print("using aiohttp : %.2fs" % (t_end_2-t_start_2))

ratio = (t_end_1-t_start_1)/(t_end_2-t_start_2)
print("ratio : %.2f" % ratio)

So is that biased ? Are there any ways to improve it to be more reliable ? Should I also monitor CPU and/or RAM usage ? anything else I'm missing ? Are there ways to improve this ?

来源：https://stackoverflow.com/questions/50030865/is-that-benchmark-reliable-aiohttp-vs-requests