之前写过一个FunTester框架对Redis进行几种常见API进行性能测试的系列文章,今天如约开始写FunTester框架对MySQL常用功能的性能测试。
我们来回顾一些Redis性能测试的文章:
- FunTester框架Redis压测预备
- FunTester框架Redis性能测试实践
- FunTester框架Redis性能测试之list操作
- FunTester框架Redis性能测试之map&INCR
MySQL系列文章预计分两期,分别是增删、改查,进行简单的SQL语句操作,如在实际工作中用到,需要自行实现SQL语句的参数化,满足业务需求。
MySQL是我本地的服务,用的也是本机,本次测试只演示常见增删改查功能的性能测试,不测试MySQL服务性能,也不设置索引等参数。后续等我学完Go语言的gorm框架,我再写一篇Go语言的MySQL性能测试文章分享。
准备工作省略不写,无非就是起个服务,建个库表,主要name和age两个字段,一个字符串一个整型数字。
insert
这个在工作中用到的不少,估计在造数据的时候用到会比较多。这里使用单条数据执行的方式。
脚本用到SQL:INSERT INTO funtesters (name,age) VALUES (\"fun${StringUtil.getString(10)}\",${getRandomInt(100)});
翻译成Java:"INSERT INTO funtesters (name,age) VALUES (\"fun"+StringUtil.getString(10)+"\","+getRandomInt(100)+");"
参数化:通过随机字符串和随机数字构成,由于本项目没有去重需求,所以没有采取唯一变量的方式。
脚本用例
/**
* MySQL insert语句实践
*/
class MysqlInsert extends SqlBase {
static final String url = "jdbc:mysql://localhost:3306/funtester?useUnicode=true&characterEncoding=utf-8&useOldAliasMetadataBehavior=true&useSSL=false"
static final int thread = 20
static final int times = 10000
public static void main(String[] args) {
RUNUP_TIME = 0
def task = []
thread.times {
task << new FunTester()
}
new Concurrent(task, "FunTester框架测试MySQL").start()
FunLibrary.testOver()
}
private static class FunTester extends FixedThread {
def connection = getConnection(url, "root", "root123456")
def statement = getStatement(connection)
FunTester() {
super(null, times, true)
}
@Override
protected void doing() throws Exception {
statement.execute("INSERT INTO funtesters (name,age) VALUES (\"fun${StringUtil.getString(10)}\",${getRandomInt(100)});")
}
@Override
protected void after() {
super.after()
close(connection, statement)
}
@Override
FixedThread clone() {
return new FunTester(limit)
}
}
}
测试结果
~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~ JSON ~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~
> {
> ① . "rt":1,
> ① . "failRate":0.0,
> ① . "threads":20,
> ① . "deviation":"13.1%",
> ① . "errorRate":0.0,
> ① . "executeTotal":198907,
> ① . "qps2":17380.898287312128,
> ① . "total":198907,
> ① . "qps":20000.0,
> ① . "startTime":"2021-11-17 15:32:52",
> ① . "endTime":"2021-11-17 15:33:03",
> ① . "mark":"FunTester框架测试MySQL171532",
> ① . "table":"eJzj5VIgCNxK80JSi0tSi54tbHs2b9uzrd0v1k/1rQwO9FEwMlAoyShKTUwhbAovFy9+u4JSiwvy84pTFUIyc1OtFCp0i1OLMhNzFPJKc3UUKnVzU1MyE/MI2UHYHQoKuZl5ChDTrAwUcot1chMrrIyADGL0EvQFOeDRtA4gwkuN2jpq66ito7aO2jpq68iwlQilQ4wa9dnQo0Z9NvSoUZ8NPWrUZ0OPGvXZ0KNGfTb0qFGfDT1qOPsMAApg90I="
> }
~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~ JSON ~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~
看QPS还挺快的,哈哈。
delete
删除这个比较麻烦,因为这些数据删了就没了。如果采取如何对消息队列做性能测试这个方案,有点大材小用,我这里就用随机名称作为删除数据的依据,执行先insert再delete的测试方案。
测试用例
/**
* MySQL insert语句实践
*/
class MysqlDelete extends SqlBase {
static final String url = "jdbc:mysql://localhost:3306/funtester?useUnicode=true&characterEncoding=utf-8&useOldAliasMetadataBehavior=true&useSSL=false"
static final int thread = 10
static final int times = 100
public static void main(String[] args) {
RUNUP_TIME = 0
def task = []
thread.times {
task << new FunTester()
}
new Concurrent(task, "FunTester框架测试MySQL").start()
FunLibrary.testOver()
}
private static class FunTester extends FixedThread {
def connection = getConnection(url, "root", "root123456")
def statement = getStatement(connection)
FunTester() {
super(null, times, true)
}
@Override
protected void doing() throws Exception {
String name = "fun${StringUtil.getString(10)}"
statement.execute("INSERT INTO funtesters (name,age) VALUES (\"$name\",${getRandomInt(100)});")
statement.execute("DELETE FROM funtesters WHERE name = \"$name\";")
}
@Override
protected void after() {
super.after()
close(connection, statement)
}
@Override
FixedThread clone() {
return new FunTester(limit)
}
}
}
测试结果
~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~ JSON ~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~
> {
> ① . "rt":1354,
> ① . "failRate":0.0,
> ① . "threads":10,
> ① . "deviation":"0.47%",
> ① . "errorRate":0.0,
> ① . "executeTotal":1000,
> ① . "qps2":7.3508872520913275,
> ① . "total":1000,
> ① . "qps":7.385524372230428,
> ① . "startTime":"2021-11-17 15:51:41",
> ① . "endTime":"2021-11-17 15:53:57",
> ① . "mark":"FunTester框架测试MySQL171551",
> ① . "table":"eJwBLwDQ/+aVsOaNrumHj+WkquWwkSzml6Dms5Xnu5jlm74hIOW6lOW9k+Wkp+S6jiAxMDI0/eodgA=="
> }
~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~ JSON ~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~~☢~
可以看到,受delete方法的影响,性能下降挺多的。
敬请期待下期的select和update以及未来可能的Go语言实践,Java和Go的HTTP客户端和服务端对比压测已经在路上了。