gevent_test.py

import datetime
import gevent
from gevent import wsgi, pywsgi

# Configuration bits.
PORT = 8081
LOG_PATH = '/Users/daniellindsley/Desktop/LoggingFramework/benchmarks/receiver/logs/gevent_test.log'

# ============

def write_to_log(env, status_code=200):
    log = open(LOG_PATH, 'a')
    now = datetime.datetime.now()
    message = "%s - [%s] %s\n" % (now, status_code, env['PATH_INFO'])
    log.write(message)
    log.close()

def error_404(env, start_response):
    gevent.spawn(write_to_log, env, 404)
    start_response('404 Not Found', [('Content-Type', 'text/html')])
    return ['Sorry, you must be thinking of something else.']

def process_request(env, start_response):
    gevent.spawn(write_to_log, env, 200)
    start_response('200 OK', [('Content-Type', 'text/html')])
    return ['OK']

def handle_request(env, start_response):
    if env['PATH_INFO'] == '/':
        return process_request(env, start_response)
    else:
        return error_404(env, start_response)

# Fast but non-streaming?
# server = wsgi.WSGIServer(('0.0.0.0', PORT), handle_request)
# Slower but streams responses.
server = pywsgi.WSGIServer(('0.0.0.0', PORT), handle_request, log=open('/dev/null', 'w'))
print('gevent_test listening on http://localhost:%s' % PORT)
server.serve_forever()

node_test.js

// Configuration bits.
var PORT = 8080;
var LOG_PATH = '/Users/travis/node_test.log';

// =============

var fs = require('fs');
var http = require('http');
var sys = require('sys');
var puts = sys.puts;

function write_to_log_sync(request, status_code) {
  var log = fs.createWriteStream(LOG_PATH, {'flags': 'a'});
  var current_time = new Date();
  var message = "" + current_time + " - [" + status_code + "] " + request.url + "\n";
  log.write(message);
  log.end();
}


function write_to_log(request, status_code) {
    fs.open(LOG_PATH, "a", 0666, function(err, fd) {
        if (err) {
            sys.puts("ERROR?");
            return;
        }
        var current_time = new Date();
        var message = "" + current_time + " - [" + status_code + "] " + request.url + "\n";
        fs.write(fd, message);
        fs.close(fd);
    });
}

function error_404(request, response) {
  write_to_log(request, 404);
  response.writeHead(404, {'Content-Type': 'text/html'});
  response.write('Sorry, you must be thinking of something else.');
  response.end();
}

function process_request(request, response) {
  write_to_log(request, 200);
  response.writeHead(200, {'Content-Type': 'text/html'});
  response.write('OK');
  response.end();
}

http.createServer(function(request, response) {
  if(request.url == '/') {
    process_request(request, response);
  }
  else {
    error_404(request, response);
  }
}).listen(PORT);
sys.puts('node_test listening on http://localhost:'+PORT);

node_vs_gevent.rst

Node vs. Gevent

Specs

• 2008 MacBook Pro (non-unibody)
• 2.4 Ghz Core 2 Duo
• 2 Gb of RAM

How

Fresh reboot Running only:

• TextMate
• Terminal.app w/ 2 tabs
• Activity Monitor (to watch CPU use & memory)

Each test was run 3-4 times, with the middle numbers of the test reported. Not a proper average.

Notes

• The gevent_test is using gevent.pywsgi instead of gevent.wsgi. Slower but supports streaming to be equivalent to Node's streaming by default.
• Node is writing to the file in an async manner, while gevent have the file write in a coroutine (might be blocking).

Node

Unloaded RAM: 5.91 Mb

ab options	Failed reqs	Reqs per sec	Secs per req	Secs per req (concurrent)	Max Mb
-c 10 -n 1000	0	4298.28	2.327	0.233	10.18
-c 25 -n 1000	0	4432.60	5.640	0.226	11.38
-c 50 -n 1000	0	4582.74	10.910	0.218	12.23
-c 100 -n 1000	0	4084.32	24.484	0.245	12.86
-c 200 -n 1000	0	3819.53	52.362	0.262	13.70
-c 300 -n 1000	0	3282.97	91.381	0.305	14.20
-c 400 -n 1000	0	3367.26	118.791	0.297	14.50
-c 500 -n 1000	0	3402.62	146.946	0.294	14.89

Gevent

Unloaded RAM: 8.52 Mb

ab options	Failed reqs	Reqs per sec	Secs per req	Secs per req (concurrent)	Max Mb
-c 10 -n 1000	0	1501.93	6.658	0.666	8.60
-c 25 -n 1000	0	1300.63	19.221	0.769	8.60?
-c 50 -n 1000	0	1418.63	35.245	0.705	8.60?
-c 100 -n 1000	0	1329.82	75.198	0.752	8.60?
-c 200 -n 1000	0	1223.89	163.414	0.817	8.60?
-c 300 -n 1000	0	1145.65	275.339	0.918	8.60?
-c 400 -n 1000	0[1]	1111.43	359.897	0.900	8.60?
-c 500 -n 1000	N/A[2]	N/A	N/A	N/A	N/A