MyBatis(十)：Mybatis 几种批量操作的对比

本章主要讲解几种批量处理的用法及对别，批量处理一般用法包含以下几种：

1）普通foreach处理（没循环一次执行一次与mysql服务器交互操作），实际上也是采用的ExecutorType.SIMPLE；

2）使用ExecutorType.BATCH批量处理方法；

3）拼接SQL，一次批量提交给Mysql多个插入语句到mysql服务器端，执行批量操作。

下边针对这几种方案分别进行示例展示用法，以及优缺点对比。

新建maven项目具体工作参考：《MyBatis(九)：Mybatis Java API批量操作（增、删、改、查） 》
在spring-config.xml中添加sqlSessionTemplate bean:

   <bean id="sqlSessionTemplate" class="org.mybatis.spring.SqlSessionTemplate">
       <constructor-arg index="0" ref="sqlSessionFactory" />
       <!--<constructor-arg index="1" value="BATCH" />-->
   </bean>

新建测试类：com.dx.test.TestBatchInsert.java

import java.util.ArrayList;
import java.util.Date;
import java.util.List;
import java.util.Random;

import org.apache.ibatis.session.ExecutorType;
import org.apache.ibatis.session.SqlSession;
import org.junit.Test;
import org.junit.runner.RunWith;
import org.mybatis.spring.SqlSessionTemplate;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.test.context.ContextConfiguration;
import org.springframework.test.context.junit4.SpringJUnit4ClassRunner;

import com.dx.test.mapper.ArticleCategoryMapper;
import com.dx.test.model.ArticleCategory;
import com.dx.test.model.enums.DataStatus;

@RunWith(SpringJUnit4ClassRunner.class)
@ContextConfiguration({ "classpath:spring-config.xml" })
public class TestBatchInsert {
   @Autowired
   private SqlSessionTemplate sqlSessionTemplate; // 引入注入的sqlSessionTemplate bean实例
   @Autowired
   private ArticleCategoryMapper articleCategoryMapper; // 参考上篇文章该类的定义。

}

mybatis 批量处理测试：

方案1）采用mybatis foreach循环插入方案：

   @Test
   public void testWithSimple() {
      long start = System.currentTimeMillis();
      SqlSession sqlSession = sqlSessionTemplate.getSqlSessionFactory().openSession(ExecutorType.SIMPLE, false);
      ArticleCategoryMapper articleCategoryMapper=sqlSession.getMapper(ArticleCategoryMapper.class);

      List<ArticleCategory> list = new ArrayList<>();
      for (int i = 0; i < 10000; i++) {
         ArticleCategory articleCategory = new ArticleCategory();
         articleCategory.setTitle("title_" + new Random().nextInt(1000) + "_" + new Random().nextInt(1000));
         articleCategory.setDescription("category description");
         articleCategory.setImgSrc("http://www.test.com/img/category/img-" + new Random().nextInt(1000) + ".gif");
         articleCategory.setState(DataStatus.Living);
         articleCategory.setCreateTime(new Date());
         articleCategory.setCreateUser("create user");
         articleCategory.setCreateUserId("user-" + new Random().nextInt(1000));

         articleCategoryMapper.insert(articleCategory);
      }
      sqlSession.commit();
      sqlSession.clearCache();
      sqlSession.close();

      long stop = System.currentTimeMillis();
      System.out.println("testWithExecutorType.SIMPLE:" + (stop - start) + "ms");
   }

执行该测试方法：

==>  Preparing: INSERT INTO article_category (title, img_src, description, state, create_user, create_user_id, create_time, update_user, update_user_id, update_time, version) VALUES (?, ?, ?, ?, ?, ?, now(), ?, ?, now(), 0)
==> Parameters: title_159_754(String), http://www.test.com/img/category/img-606.gif(String), category description(String), 0(Integer), create user(String), user-740(String), null, null
<==    Updates: 1
。。。
==>  Preparing: INSERT INTO article_category (title, img_src, description, state, create_user, create_user_id, create_time, update_user, update_user_id, update_time, version) VALUES (?, ?, ?, ?, ?, ?, now(), ?, ?, now(), 0)
==> Parameters: title_416_561(String), http://www.test.com/img/category/img-72.gif(String), category description(String), 0(Integer), create user(String), user-98(String), null, null
<==    Updates: 1
==>  Preparing: INSERT INTO article_category (title, img_src, description, state, create_user, create_user_id, create_time, update_user, update_user_id, update_time, version) VALUES (?, ?, ?, ?, ?, ?, now(), ?, ?, now(), 0)
==> Parameters: title_967_656(String), http://www.test.com/img/category/img-897.gif(String), category description(String), 0(Integer), create user(String), user-620(String), null, null
<==    Updates: 1
==>  Preparing: INSERT INTO article_category (title, img_src, description, state, create_user, create_user_id, create_time, update_user, update_user_id, update_time, version) VALUES (?, ?, ?, ?, ?, ?, now(), ?, ?, now(), 0)
==> Parameters: title_399_676(String), http://www.test.com/img/category/img-819.gif(String), category description(String), 0(Integer), create user(String), user-889(String), null, null
<==    Updates: 1
testWithExecutorType.SIMPLE::8163ms

测试结果发现耗时为：8163ms，而且从打印日志，可以看出该类执行是一条一条的插入的，而且每次插入前都要Preparing插入sql，这个是比较耗时的。

方案2）采用mybatis batch方案（使用ExecutorType.BATCH）：

   @Test
   public void testWithBatch() {
      long start = System.currentTimeMillis();
      SqlSession sqlSession = sqlSessionTemplate.getSqlSessionFactory().openSession(ExecutorType.BATCH, false);
      ArticleCategoryMapper articleCategoryMapper=sqlSession.getMapper(ArticleCategoryMapper.class);
      List<ArticleCategory> list = new ArrayList<>();
      for (int i = 0; i < 10000; i++) {
         ArticleCategory articleCategory = new ArticleCategory();
         articleCategory.setTitle("title_" + new Random().nextInt(1000) + "_" + new Random().nextInt(1000));
         articleCategory.setDescription("category description");
         articleCategory.setImgSrc("http://www.test.com/img/category/img-" + new Random().nextInt(1000) + ".gif");
         articleCategory.setState(DataStatus.Living);
         articleCategory.setCreateTime(new Date());
         articleCategory.setCreateUser("create user");
         articleCategory.setCreateUserId("user-" + new Random().nextInt(1000));

         articleCategoryMapper.insert(articleCategory);
      }
      sqlSession.commit();
      sqlSession.clearCache();
      sqlSession.close();

      long stop = System.currentTimeMillis();
      System.out.println("testWithExecutorType.BATCH:" + (stop - start) + "ms");
   }

执行该测试方法：

==>  Preparing: INSERT INTO article_category (title, img_src, description, state, create_user, create_user_id, create_time, update_user, update_user_id, update_time, version) VALUES (?, ?, ?, ?, ?, ?, now(), ?, ?, now(), 0)
==> Parameters: title_345_236(String), http://www.test.com/img/category/img-688.gif(String), category description(String), 0(Integer), create user(String), user-337(String), null, null
。。。
==> Parameters: title_534_211(String), http://www.test.com/img/category/img-572.gif(String), category description(String), 0(Integer), create user(String), user-139(String), null, null
==> Parameters: title_292_271(String), http://www.test.com/img/category/img-8.gif(String), category description(String), 0(Integer), create user(String), user-547(String), null, null
testWithExecutorType.BATCH:5283ms

测试结果发现耗时为：5283ms，而且从打印日志可以看出该方案只一致性一次预编译，之后插入都不需要再次预编译。

方案3）采用拼接SQL方案：

用法1：

   @Test
   public void testWithBatchScript() {
      long start = System.currentTimeMillis();

      List<ArticleCategory> list = new ArrayList<>();
      for (int i = 0; i < 10000; i++) {
         ArticleCategory articleCategory = new ArticleCategory();
         articleCategory.setTitle("title_" + new Random().nextInt(1000) + "_" + new Random().nextInt(1000));
         articleCategory.setDescription("category description");
         articleCategory.setImgSrc("http://www.test.com/img/category/img-" + new Random().nextInt(1000) + ".gif");
         articleCategory.setState(DataStatus.Living);
         articleCategory.setCreateTime(new Date());
         articleCategory.setCreateUser("create user");
         articleCategory.setCreateUserId("user-" + new Random().nextInt(1000));

         list.add(articleCategory);
      }
      articleCategoryMapper.batchInsertsWithScript(list);

      long stop = System.currentTimeMillis();
      System.out.println("testWithBatchScript:" + (stop - start) + "ms");
   }

该方法执行日志：

JDBC Connection [com.alibaba.druid.proxy.jdbc.ConnectionProxyImpl@1b84f475] will not be managed by Spring
==>  Preparing: INSERT INTO `article_category` (`title`,`img_src`,`description`,`state`,`create_time`,`create_user`,`create_user_id`,`update_time`,`update_user`,`update_user_id`,`version`)
VALUES (?,?,?,?,now(),?,?,now(),?,?,0) ,
。。。
(?,?,?,?,now(),?,?,now(),?,?,0) ,
(?,?,?,?,now(),?,?,now(),?,?,0)
ON DUPLICATE KEY UPDATE `update_time` = now()
==> Parameters:
title_375_245(String), http://www.test.com/img/category/img-247.gif(String), category description(String), 0(Integer), create user(String), user-930(String), null, null, 
。。。
title_275_39(String), http://www.test.com/img/category/img-875.gif(String), category description(String), 0(Integer), create user(String), user-323(String), null, null, 
title_743_735(String), http://www.test.com/img/category/img-220.gif(String), category description(String), 0(Integer), create user(String), user-917(String), null, null
<==    Updates: 10000
Closing non transactional SqlSession [org.apache.ibatis.session.defaults.DefaultSqlSession@65b104b9]
testWithBatchScript:3312ms

测试2：

   @Test
   public void testWithBatchSQL() {
      long start = System.currentTimeMillis();

      List<ArticleCategory> list = new ArrayList<>();
      for (int i = 0; i < 10000; i++) {
         ArticleCategory articleCategory = new ArticleCategory();
         articleCategory.setTitle("title_" + new Random().nextInt(1000) + "_" + new Random().nextInt(1000));
         articleCategory.setDescription("category description");
         articleCategory.setImgSrc("http://www.test.com/img/category/img-" + new Random().nextInt(1000) + ".gif");
         articleCategory.setState(DataStatus.Living);
         articleCategory.setCreateTime(new Date());
         articleCategory.setCreateUser("create user");
         articleCategory.setCreateUserId("user-" + new Random().nextInt(1000));

         list.add(articleCategory);
      }
      articleCategoryMapper.batchInserts(list);
      long stop = System.currentTimeMillis();
      System.out.println("testWithBatchSQL:" + (stop - start) + "ms");
   }

执行该方法打印日志：

Creating a new SqlSession
SqlSession [org.apache.ibatis.session.defaults.DefaultSqlSession@65b104b9] was not registered for synchronization because synchronization is not active
JDBC Connection [com.alibaba.druid.proxy.jdbc.ConnectionProxyImpl@31ff43be] will not be managed by Spring
==>  Preparing: INSERT INTO `article_category`(`title`,`img_src`,`description`,`state`,`create_time`,`create_user`,`create_user_id`,`update_time`,`update_user`,`update_user_id`,`version`)
VALUES(?,?,?,?,now(),?,?,now(),?,?,0),
。。。
(?,?,?,?,now(),?,?,now(),?,?,0),
(?,?,?,?,now(),?,?,now(),?,?,0)
ON DUPLICATE KEY UPDATE `update_time` = now();
==> Parameters: title_131_603(String), http://www.test.com/img/category/img-831.gif(String), category description(String), 0(Integer), create user(String), user-35(String), null, null, 
。。。
title_317_725(String), http://www.test.com/img/category/img-208.gif(String), category description(String), 0(Integer), create user(String), user-968(String), null, null, 
title_403_241(String), http://www.test.com/img/category/img-870.gif(String), category description(String), 0(Integer), create user(String), user-483(String), null, null
<==    Updates: 10000
Closing non transactional SqlSession [org.apache.ibatis.session.defaults.DefaultSqlSession@65b104b9]
testWithBatchSQL:2978ms

上边这两种方法实际上原理相同：拼接SQL，一次提交一批SQL方案。

拼接SQL一次提交方案缺点：当拼接sql超出’max_allowed_packet‘设置大小时，会抛出异常：

com.mysql.jdbc.PacketTooBigException: Packet for query is too large (8346602 > 4194304（4M）). You can change this value on the server by setting the max_allowed_packet’ variable.

解决错误方案：

我安装的mysql版本是：8.0.17 MySQL Community Server - GPL

dx:~ $ mysql -uroot -p123456 -h127.0.0.1
mysql: [Warning] Using a password on the command line interface can be insecure.
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 356
Server version: 8.0.17 MySQL Community Server - GPL

Copyright (c) 2000, 2019, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

该版本的mysql允许的发送给服务器的最大数据包大小情况如下：

ysql> show variables like '%packet%';
+---------------------------+------------+
| Variable_name             | Value      |
+---------------------------+------------+
| max_allowed_packet        | 67108864   |
| mysqlx_max_allowed_packet | 67108864   |
| slave_max_allowed_packet  | 1073741824 |
+---------------------------+------------+
3 rows in set (0.00 sec)

ax_allowed_packet 设置以字节发送给服务器的最大数据包大小. (默认: 64MB)，
数据库变更需要：大的插入和更新可能会被max_allowed_packet参数限制，导致失败。
此时，解决方案：
1）在mysql中执行：set global max_allowed_paclet = 128*1024*1024；
2）修改mysql服务器的配置(my.ini 或者 my.cnf 文件)参数： max_allowed_packet = 128M 来解决，
3）需要重启mysql服务器（缺陷）。

性能总结：

+----------------------------------------------+------------+
| 采用方案                                      | 耗时        |
+----------------------------------------------+------------+
| 普通处理方法(ExecutorType.SIMPLE)逐条插入处理    | 8s         |
| 使用ExecutorType.BATCH                        | 5s         |
| 使用拼接SQL,一致性提交一批SQL                    | 3s         |
+-----------------------------------------------+------------+