性能优化之Java多线程批量拆分List导入数据库

た入场券 2022-09-15 05:54 216阅读 0赞

点击上方蓝色“石杉的架构笔记”，选择“设为星标”

回复“PDF”获取独家整理的学习资料！

![5248f6d4b1649414c41c8c97370a492c.png][]

![e1aab691c4d23080aa670a0740a41783.png][]

![c8af9da1fbb76b0a96e9fee2b0b83ad6.png][]

长按扫描上方一元购买

![386131c44c3cababd764916754ae4a92.png][]

来源：https://blog.csdn.net/kisscatforever/article/details/79817039

## | 前言 ##

前两天做了一个导入的功能，导入开始的时候非常慢，导入2w条数据要1分多钟，后来一点一点的优化，从直接把list怼进Mysql中，到分配把list导入Mysql中，到多线程把list导入Mysql中。时间是一点一点的变少了。非常的爽，最后变成了10s以内。下面就展示一下过程。

## | 直接把list怼进Mysql ##

使用mybatis的批量导入操作：

@Transactional(rollbackFor = Exception.class)
        public int addFreshStudentsNew2(List<FreshStudentAndStudentModel> list, String schoolNo) {
            if (list == null || list.isEmpty()) {
                return 0;
            }
            List<StudentEntity> studentEntityList = new LinkedList<>();
            List<EnrollStudentEntity> enrollStudentEntityList = new LinkedList<>();
            List<AllusersEntity> allusersEntityList = new LinkedList<>();
    
            for (FreshStudentAndStudentModel freshStudentAndStudentModel : list) {
    
                EnrollStudentEntity enrollStudentEntity = new EnrollStudentEntity();
                StudentEntity studentEntity = new StudentEntity();
                BeanUtils.copyProperties(freshStudentAndStudentModel, studentEntity);
                BeanUtils.copyProperties(freshStudentAndStudentModel, enrollStudentEntity);
                String operator = TenancyContext.UserID.get();
                String studentId = BaseUuidUtils.base58Uuid();
                enrollStudentEntity.setId(BaseUuidUtils.base58Uuid());
                enrollStudentEntity.setStudentId(studentId);
                enrollStudentEntity.setIdentityCardId(freshStudentAndStudentModel.getIdCard());
                enrollStudentEntity.setOperator(operator);
                studentEntity.setId(studentId);
                studentEntity.setIdentityCardId(freshStudentAndStudentModel.getIdCard());
                studentEntity.setOperator(operator);
                studentEntityList.add(studentEntity);
                enrollStudentEntityList.add(enrollStudentEntity);
    
                AllusersEntity allusersEntity = new AllusersEntity();
                allusersEntity.setId(enrollStudentEntity.getId());
                allusersEntity.setUserCode(enrollStudentEntity.getNemtCode());
                allusersEntity.setUserName(enrollStudentEntity.getName());
                allusersEntity.setSchoolNo(schoolNo);
                allusersEntity.setTelNum(enrollStudentEntity.getTelNum());
                allusersEntity.setPassword(enrollStudentEntity.getNemtCode());  //密码设置为考生号
                allusersEntityList.add(allusersEntity);
            }
                enResult = enrollStudentDao.insertAll(enrollStudentEntityList);
                stuResult = studentDao.insertAll(studentEntityList);
                allResult = allusersFacade.insertUserList(allusersEntityList);
    
            if (enResult > 0 && stuResult > 0 && allResult) {
                return 10;
            }
            return -10;
        }

Mapper.xml

<insert id="insertAll" parameterType="com.dmsdbj.itoo.basicInfo.entity.EnrollStudentEntity">
            insert into tb_enroll_student
            <trim prefix="(" suffix=")" suffixOverrides=",">
                    id,  
                    remark,  
                    nEMT_aspiration,  
                    nEMT_code,  
                    nEMT_score,  
                    student_id,  
                    identity_card_id,  
                    level,  
                    major,  
                    name,  
                    nation,  
                    secondary_college,  
                    operator,  
                    sex,  
                    is_delete,  
                    account_address,  
                    native_place,  
                    original_place,  
                    used_name,  
                    pictrue,  
                    join_party_date,  
                    political_status,  
                    tel_num,  
                    is_registry,  
                    graduate_school,  
                    create_time,  
                    update_time        </trim>        
            values
            <foreach collection="list" item="item" index="index" separator=",">
            (
                    #{item.id,jdbcType=VARCHAR},
                    #{item.remark,jdbcType=VARCHAR},
                    #{item.nemtAspiration,jdbcType=VARCHAR},
                    #{item.nemtCode,jdbcType=VARCHAR},
                    #{item.nemtScore,jdbcType=VARCHAR},
                    #{item.studentId,jdbcType=VARCHAR},
                    #{item.identityCardId,jdbcType=VARCHAR},
                    #{item.level,jdbcType=VARCHAR},
                    #{item.major,jdbcType=VARCHAR},
                    #{item.name,jdbcType=VARCHAR},
                    #{item.nation,jdbcType=VARCHAR},
                    #{item.secondaryCollege,jdbcType=VARCHAR},
                    #{item.operator,jdbcType=VARCHAR},
                    #{item.sex,jdbcType=VARCHAR},
                    0,
                    #{item.accountAddress,jdbcType=VARCHAR},
                    #{item.nativePlace,jdbcType=VARCHAR},
                    #{item.originalPlace,jdbcType=VARCHAR},
                    #{item.usedName,jdbcType=VARCHAR},
                    #{item.pictrue,jdbcType=VARCHAR},
                    #{item.joinPartyDate,jdbcType=VARCHAR},
                    #{item.politicalStatus,jdbcType=VARCHAR},
                    #{item.telNum,jdbcType=VARCHAR},
                    #{item.isRegistry,jdbcType=TINYINT},
                    #{item.graduateSchool,jdbcType=VARCHAR},
                    now(),
                    now()        
            )   
            </foreach>                
      </insert>

代码说明：

底层的mapper是通过逆向工程来生成的，批量插入如下，是拼接成类似：`insert into tb_enroll_student()values ()，()…….();`

这样的缺点是，数据库一般有一个默认的设置，就是每次sql操作的数据不能超过4M。这样插入，数据多的时候，数据库会报错`Packet for query is too large (6071393 > 4194304). You can change this value on the server by setting the max_allowed_packet' variable.，`虽然我们可以通过

类似 修改 my.ini 加上 `max_allowed_packet =67108864`，`67108864=64M`，默认大小4194304 也就是4M

修改完成之后要重启mysql服务，如果通过命令行修改就不用重启mysql服务。

完成本次操作，但是我们不能保证项目单次最大的大小是多少，这样是有弊端的。所以可以考虑进行分组导入。

## | 分组把list导入Mysql中 ##

同样适用mybatis批量插入，区别是对每次的导入进行分组计算，然后分多次进行导入：

@Transactional(rollbackFor = Exception.class)
        public int addFreshStudentsNew2(List<FreshStudentAndStudentModel> list, String schoolNo) {
            if (list == null || list.isEmpty()) {
                return 0;
            }
            List<StudentEntity> studentEntityList = new LinkedList<>();
            List<EnrollStudentEntity> enrollStudentEntityList = new LinkedList<>();
            List<AllusersEntity> allusersEntityList = new LinkedList<>();
    
            for (FreshStudentAndStudentModel freshStudentAndStudentModel : list) {
    
                EnrollStudentEntity enrollStudentEntity = new EnrollStudentEntity();
                StudentEntity studentEntity = new StudentEntity();
                BeanUtils.copyProperties(freshStudentAndStudentModel, studentEntity);
                BeanUtils.copyProperties(freshStudentAndStudentModel, enrollStudentEntity);
                String operator = TenancyContext.UserID.get();
                String studentId = BaseUuidUtils.base58Uuid();
                enrollStudentEntity.setId(BaseUuidUtils.base58Uuid());
                enrollStudentEntity.setStudentId(studentId);
                enrollStudentEntity.setIdentityCardId(freshStudentAndStudentModel.getIdCard());
                enrollStudentEntity.setOperator(operator);
                studentEntity.setId(studentId);
                studentEntity.setIdentityCardId(freshStudentAndStudentModel.getIdCard());
                studentEntity.setOperator(operator);
                studentEntityList.add(studentEntity);
                enrollStudentEntityList.add(enrollStudentEntity);
    
                AllusersEntity allusersEntity = new AllusersEntity();
                allusersEntity.setId(enrollStudentEntity.getId());
                allusersEntity.setUserCode(enrollStudentEntity.getNemtCode());
                allusersEntity.setUserName(enrollStudentEntity.getName());
                allusersEntity.setSchoolNo(schoolNo);
                allusersEntity.setTelNum(enrollStudentEntity.getTelNum());
                allusersEntity.setPassword(enrollStudentEntity.getNemtCode());  //密码设置为考生号
                allusersEntityList.add(allusersEntity);
            }
    
            int c = 100;
            int b = enrollStudentEntityList.size() / c;
            int d = enrollStudentEntityList.size() % c;
    
            int enResult = 0;
            int stuResult = 0;
            boolean allResult = false;
    
            for (int e = c; e <= c * b; e = e + c) {
                enResult = enrollStudentDao.insertAll(enrollStudentEntityList.subList(e - c, e));
                stuResult = studentDao.insertAll(studentEntityList.subList(e - c, e));
                allResult = allusersFacade.insertUserList(allusersEntityList.subList(e - c, e));
            }
            if (d != 0) {
                enResult = enrollStudentDao.insertAll(enrollStudentEntityList.subList(c * b, enrollStudentEntityList.size()));
                stuResult = studentDao.insertAll(studentEntityList.subList(c * b, studentEntityList.size()));
                allResult = allusersFacade.insertUserList(allusersEntityList.subList(c * b, allusersEntityList.size()));
            }
    
            if (enResult > 0 && stuResult > 0 && allResult) {
                return 10;
            }
            return -10;
        }

代码说明：

这样操作，可以避免上面的错误，但是分多次插入，无形中就增加了操作实践，很容易超时。所以这种方法还是不值得提倡的。

再次改进，使用多线程分批导入。

## | 多线程分批导入Mysql ##

依然使用mybatis的批量导入，不同的是，根据线程数目进行分组，然后再建立多线程池，进行导入。

@Transactional(rollbackFor = Exception.class)
        public int addFreshStudentsNew(List<FreshStudentAndStudentModel> list, String schoolNo) {
            if (list == null || list.isEmpty()) {
                return 0;
            }
            List<StudentEntity> studentEntityList = new LinkedList<>();
            List<EnrollStudentEntity> enrollStudentEntityList = new LinkedList<>();
            List<AllusersEntity> allusersEntityList = new LinkedList<>();
    
            list.forEach(freshStudentAndStudentModel -> {
                EnrollStudentEntity enrollStudentEntity = new EnrollStudentEntity();
                StudentEntity studentEntity = new StudentEntity();
                BeanUtils.copyProperties(freshStudentAndStudentModel, studentEntity);
                BeanUtils.copyProperties(freshStudentAndStudentModel, enrollStudentEntity);
                String operator = TenancyContext.UserID.get();
                String studentId = BaseUuidUtils.base58Uuid();
                enrollStudentEntity.setId(BaseUuidUtils.base58Uuid());
                enrollStudentEntity.setStudentId(studentId);
                enrollStudentEntity.setIdentityCardId(freshStudentAndStudentModel.getIdCard());
                enrollStudentEntity.setOperator(operator);
                studentEntity.setId(studentId);
                studentEntity.setIdentityCardId(freshStudentAndStudentModel.getIdCard());
                studentEntity.setOperator(operator);
                studentEntityList.add(studentEntity);
                enrollStudentEntityList.add(enrollStudentEntity);
    
                AllusersEntity allusersEntity = new AllusersEntity();
                allusersEntity.setId(enrollStudentEntity.getId());
                allusersEntity.setUserCode(enrollStudentEntity.getNemtCode());
                allusersEntity.setUserName(enrollStudentEntity.getName());
                allusersEntity.setSchoolNo(schoolNo);
                allusersEntity.setTelNum(enrollStudentEntity.getTelNum());
                allusersEntity.setPassword(enrollStudentEntity.getNemtCode());  //密码设置为考生号
                allusersEntityList.add(allusersEntity);
            });
    
    
            int nThreads = 50;
    
            int size = enrollStudentEntityList.size();
            ExecutorService executorService = Executors.newFixedThreadPool(nThreads);
            List<Future<Integer>> futures = new ArrayList<Future<Integer>>(nThreads);
    
            for (int i = 0; i < nThreads; i++) {
                final List<EnrollStudentEntity> EnrollStudentEntityImputList = enrollStudentEntityList.subList(size / nThreads * i, size / nThreads * (i + 1));
                final List<StudentEntity> studentEntityImportList = studentEntityList.subList(size / nThreads * i, size / nThreads * (i + 1));
                final List<AllusersEntity> allusersEntityImportList = allusersEntityList.subList(size / nThreads * i, size / nThreads * (i + 1));
    
               Callable<Integer> task1 = () -> {
              studentSave.saveStudent(EnrollStudentEntityImputList,studentEntityImportList,allusersEntityImportList);
                   return 1;
                };
              futures.add(executorService.submit(task1));
            }
            executorService.shutdown();
            if (!futures.isEmpty() && futures != null) {
                return 10;
            }
            return -10;
        }

代码说明：

上面是通过应用ExecutorService 建立了固定的线程数，然后根据线程数目进行分组，批量依次导入。一方面可以缓解数据库的压力，另一个面线程数目多了，一定程度会提高程序运行的时间。缺点就是要看服务器的配置，如果配置好的话就可以开多点线程，配置差的话就开小点。

## | 小结 ##

通过使用这个操作真是不断的提高了，项目使用技巧也是不错。加油~~ 多线程哦~~

![d86975a679e642a027de6d9e5a63a629.png][]

[5248f6d4b1649414c41c8c97370a492c.png]: /images/20220828/54422af535434f11b6a18edb8f74ef8f.png
[e1aab691c4d23080aa670a0740a41783.png]: /images/20220828/63b09d5290c34548b48d206ab4af11fd.png
[c8af9da1fbb76b0a96e9fee2b0b83ad6.png]: /images/20220828/6bc5f9f142534dcd99d13ecda8fd9b3b.png
[386131c44c3cababd764916754ae4a92.png]: /images/20220828/f96b0854cec74bd99e196a60cd7b9df7.png
[d86975a679e642a027de6d9e5a63a629.png]: /images/20220828/b5219f7c029f44319a42244d9ca88138.png