使用mybatis插入数据执行效率对比,对比三种方式(测试数据库为MySQL),
1 使用 SqlSessionFactory,每一批数据执行一次提交(
未开启批处理
)
2 使用mybatis-plus框架的insert方法,for循环,每次执行一次插入
3 使用ibatis,纯sql插入
新增xml执行效率测试:xml执行时间比sql稍慢一些,50000条数据插入时间约为2000毫秒左右,平均时间是sql的2倍左右。
先贴出执行效果(数字代表执行的时间,单位毫秒):
测试代码:
//测试类
@RunWith(SpringRunner.class)
@SpringBootTest(classes = DemoApplication.class)
public class Test1 {
@Autowired
UsersMapper usersMapper;
@Autowired
SqlSessionFactory sqlSessionFactory;
public List<Users> list = new ArrayList<>();
@Before
public void getList() {
long start = System.currentTimeMillis();
Users user;
for (int i = 1; i <=50000 ; i++) {
user = new Users();
user.setId(i);
user.setName("java");
user.setAge(200);
user.setManagerId(222);
list.add(user);
}
System.out.println("拼装数据 耗时:"+(System.currentTimeMillis()-start));
System.out.println(list.size());
}
@Test
public void batchInsert() {
SqlSession sqlSession = sqlSessionFactory.openSession(ExecutorType.BATCH,false);
UsersMapper mapper = sqlSession.getMapper(UsersMapper.class);
System.out.println("batchInsert 插入开始========");
long start = System.currentTimeMillis();
for (int i = 0; i < list.size(); i++) {
mapper.insert(list.get(i));
if (i%5000==4999) {
sqlSession.flushStatements();
// sqlSession.commit();
// sqlSession.clearCache();
}
}
// sqlSession.commit();
// sqlSession.clearCache();
sqlSession.flushStatements();
System.out.println("SqlSession 批量插入耗时:"+(System.currentTimeMillis()-start));
}
@Test
public void forEachInsert() {
System.out.println("forEachInsert 插入开始========");
long start = System.currentTimeMillis();
for (int i = 0; i < list.size(); i++) {
usersMapper.insert(list.get(i));
}
System.out.println("foreach 插入耗时:"+(System.currentTimeMillis()-start));
}
@Test
public void sqlInsert() {
System.out.println("sql 插入开始========");
long start = System.currentTimeMillis();
usersMapper.sqlInsert(list);
System.out.println("sql 插入耗时:"+(System.currentTimeMillis()-start));
}
//xml批量插入
@Test
public void xmlInsert() {
System.out.println("xmlInsert 批量插入开始========");
long start = System.currentTimeMillis();
usersMapper.xmlBatchInsert(list);
System.out.println("xmlInsert 批量插入耗时:"+(System.currentTimeMillis()-start));
}
}
//sql插入相关类
@Repository
public interface UsersMapper extends BaseMapper<Users> {
@InsertProvider(type = UsersProvider.class, method = "insertListSql")
public void sqlInsert(List<Users> list);
public void xmlBatchInsert(@Param("list") List<Users> list);
}
public class UsersProvider {
public String insertListSql(List<Users> list) {
StringBuffer sqlList = new StringBuffer();
sqlList.append(" INSERT INTO users(id,name,age,manager_id) VALUES ");
for (int i = 0; i < list.size() ; i++) {
Users user = list.get(i);
sqlList.append(" (").append(user.getId()).append(",").append("'").append(user.getName()).append("',").append(user.getAge())
.append(",").append(user.getManagerId()).append(")");
if (i < list.size()-1) {
sqlList.append(",");
}
}
return sqlList.toString();
}
}
//xml 插入mapper文件
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE mapper PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN" "http://mybatis.org/dtd/mybatis-3-mapper.dtd">
<mapper namespace="com.example.demo.dao.UsersMapper">
<insert id="xmlBatchInsert">
INSERT INTO users(id,name,age,manager_id) VALUES
<foreach collection="list" item="item" index="index" separator=",">
(#{item.id},
#{item.name},
#{item.age},
#{item.managerId})
</foreach>
</insert>
</mapper>
总结:
sql插入的效率最高,sqlsession次之,mybatis框架foreach插入效率最低。
执行效率echarts图:
补充:
1、sqlsession之所以效率比较低,是因为默认情况下,mysql驱动是会忽略我们的批操作的,要想sqlsession支持批操作要开启才可以。开启后速度和xml批量插入差不多,大约2.8秒
rewriteBatchedStatements=true
2、sql插入是有sql长度限制的,配置可以查看mysql配置文件,避免sql过长导致内存溢出
windows 默认在:C:\ProgramData\MySQL\MySQL Server 5.7\my.ini
linux默认在:/etc/my.cnf