MongoDB Insert VS Save

benmark.py

# -*- coding:utf-8 -*-
# MongoDB Benmarking with Mongoengine

import os, time, multiprocessing

import pymongo
import mongoengine

DB_HOST = '127.0.0.1'
DB_PORT = 27017

class DocA(mongoengine.Document):
    field_a = mongoengine.IntField()
    field_b = mongoengine.StringField()
    field_c = mongoengine.ListField(mongoengine.StringField())
    field_d = mongoengine.DictField()

def _conect_database():
    mongoengine.connect('TestDB', host=DB_HOST, port=DB_PORT)

def _create_database():
    connection = pymongo.Connection(DB_HOST, DB_PORT)
    db = connection['TestDB']
    connection.close()

    _conect_database()

def _drop_database():
    connection = pymongo.Connection(DB_HOST, DB_PORT)
    connection.drop_database('TestDB')

def _create_by_insert(num):

    print '>>> Insert:'
    insert_begin = time.time()

    new_docs = []
    for i in xrange(num):
        doc = DocA(
                field_a = i,
                field_b = 'Doc #%d' % i,
                field_c = ['Hello', 'XingCloud', ],
                field_d = {
                        'TBBT': 'The Big Bang Theory',
                    },
            )
        new_docs.append(doc)

    DocA.objects.insert(new_docs, load_bulk=False)

    insert_end = time.time()
    print '    Eclipse: %f' % (insert_end - insert_begin)

def _create_by_save(num):

    print 'PID: %s >>> Save:' % os.getpid()
    save_begin = time.time()

    for i in xrange(num):
         doc = DocA(
                field_a = i,
                field_b = 'Doc #%d' % i,
                field_c = ['Hello', 'XingCloud', ],
                field_d = {
                        'TBBT': 'The Big Bang Theory',
                    },
            )
         doc.save()

    save_end = time.time()
    print 'PID: %s    Eclipse: %f' % (os.getpid(), save_end - save_begin)

if __name__ == '__main__':
    SAMPLE_SIZE = 16000
    PROCESS_NUM = 2

    _create_database()
    print 'Database Created ...'

    print '=== Benchmark Begin (Sample Size: %d) ===' % SAMPLE_SIZE

    _create_database()
    print 'Database Created ...'

    _create_by_insert(SAMPLE_SIZE)
    print 'Database Count: %d' % DocA.objects.count()

    _drop_database()
    print 'Test Database Droped...'

    _create_database()
    print 'Database Created ...'

    for i in xrange(PROCESS_NUM):
        multiprocessing.Process(
                target = _create_by_save,
                args = (SAMPLE_SIZE / PROCESS_NUM, ),
            ).start()

    time.sleep(10)
    print 'Database Count: %d' % DocA.objects.count()

    _drop_database()
    print 'Test Database Droped...'

    print '=== Benchmark End ==='