python环境测试MySQLdb、DBUtil、sqlobject性能
首先介绍下MySQLdb、DBUtil、sqlobject:
(1)MySQLdb 是用于Python连接Mysql数据库的接口,它实现了 Python 数据库 API 规范 V2.0,基于 MySQL C API 上建立的。除了MySQLdb外,python还可以通过oursql, PyMySQL, myconnpy等模块实现MySQL数据库操作;
(2)DBUtil中提供了几种连接池,用以提高数据库的访问性能,例如PooledDB,PesistentDB等
(3)sqlobject可以实现数据库ORM映射的第三方模块,可以以对象、实例的方式轻松操作数据库中记录。
为测试这三者的性能,简单做一个例子:50个并发访问4000条记录的单表,数据库记录如下:
测试代码如下:
1、MySQLdb的代码如下,其中在connDB()中把连接池相关代码暂时做了一个注释,去掉这个注释既可以使用连接池来创建数据库连接:
(1)DBOperator.py
import MySQLdb from stockmining.stocks.setting import LoggerFactory import connectionpool class DBOperator(object): def __init__(self): self.logger = LoggerFactory.getLogger(‘DBOperator‘) self.conn = None def connDB(self): self.conn=MySQLdb.connect(host="127.0.0.1",user="root",passwd="root",db="pystock",port=3307,charset="utf8") #当需要使用连接池的时候开启 #self.conn=connectionpool.pool.connection() return self.conn def closeDB(self): if(self.conn != None): self.conn.close() def execute(self, sql): try: if(self.conn != None): cursor = self.conn.cursor() else: raise MySQLdb.Error(‘No connection‘) n = cursor.execute(sql) return n except MySQLdb.Error,e: self.logger.error("Mysql Error %d: %s" % (e.args[0], e.args[1])) def findBySQL(self, sql): try: if(self.conn != None): cursor = self.conn.cursor() else: raise MySQLdb.Error(‘No connection‘) cursor.execute(sql) rows = cursor.fetchall() return rows except MySQLdb.Error,e: self.logger.error("Mysql Error %d: %s" % (e.args[0], e.args[1]))
(2)测试代码testMysql.py,做了50个并发,对获取到的数据库记录做了个简单遍历。
import threading import time import DBOperator def run(): operator = DBOperator() operator.connDB() starttime = time.time() sql = "select * from stock_cash_tencent" peeps = operator.findBySQL(sql) for r in peeps: pass operator.closeDB() endtime = time.time() diff = (endtime - starttime)*1000 print diff def test(): for i in range(50): threading.Thread(target = run).start() time.sleep(1) if __name__ == ‘__main__‘: test()
2、连接池相关代码:
(1)connectionpool.py
from DBUtils import PooledDB import MySQLdb import string maxconn = 30 #最大连接数 mincached = 10 #最小空闲连接 maxcached = 20 #最大空闲连接 maxshared = 30 #最大共享连接 connstring="root#root#127.0.0.1#3307#pystock#utf8" #数据库地址 dbtype = "mysql" def createConnectionPool(connstring, dbtype): db_conn = connstring.split("#"); if dbtype==‘mysql‘: try: pool = PooledDB.PooledDB(MySQLdb, user=db_conn[0],passwd=db_conn[1],host=db_conn[2],port=string.atoi(db_conn[3]),db=db_conn[4],charset=db_conn[5], mincached=mincached,maxcached=maxcached,maxshared=maxshared,maxconnections=maxconn) return pool except Exception, e: raise Exception,‘conn datasource Excepts,%s!!!(%s).‘%(db_conn[2],str(e)) return None pool = createConnectionPool(connstring, dbtype)
3、sqlobject相关代码
(1)connection.py
from sqlobject.mysql import builder conn = builder()(user=‘root‘, password=‘root‘, host=‘127.0.0.1‘, db=‘pystock‘, port=3307, charset=‘utf8‘)
(2)StockCashTencent.py对应到数据库中的表,50个并发并作了一个简单的遍历。(实际上,如果不做遍历,只做count()计算,sqlobject性能是相当高的。)
import sqlobject import time from connection import conn import threading class StockCashTencent(sqlobject.SQLObject): _connection = conn code = sqlobject.StringCol() name = sqlobject.StringCol() date = sqlobject.StringCol() main_in_cash = sqlobject.FloatCol() main_out_cash = sqlobject.FloatCol() main_net_cash = sqlobject.FloatCol() main_net_rate= sqlobject.FloatCol() private_in_cash= sqlobject.FloatCol() private_out_cash= sqlobject.FloatCol() private_net_cash= sqlobject.FloatCol() private_net_rate= sqlobject.FloatCol() total_cash= sqlobject.FloatCol() def test(): starttime = time.time() query = StockCashTencent.select(True) for result in query: pass endtime = time.time() diff = (endtime - starttime)*1000 print diff if __name__ == ‘__main__‘: for i in range(50): threading.Thread(target = test).start() time.sleep(1)
测试结果如下:
MySQLdb平均(毫秒) | 99.63999271 |
DBUtil平均(毫秒) | 97.07998276 |
sqlobject平均(毫秒) | 343.2999897 |
结论:其实就测试数据而言,MySQLdb单连接和DBUtil连接池的性能并没有很大的区别(100个并发下也相差无几),相反sqlobject虽然具有的编程上的便利性,但是却带来性能上的巨大不足,在实际中使用哪个模块就要斟酌而定了。
本文出自 “清茶一杯,风轻云淡” 博客,转载请与作者联系!
郑重声明:本站内容如果来自互联网及其他传播媒体,其版权均属原媒体及文章作者所有。转载目的在于传递更多信息及用于网络分享,并不代表本站赞同其观点和对其真实性负责,也不构成任何其他建议。