9
利用MySQL实现分布式锁,涉及到乐观锁和悲观锁的思想 - eaglelihh
source link: https://www.cnblogs.com/eaglelihh/p/16435341.html
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
对于一些并发量不是很高的场景,使用MySQL来实现乐观锁会比较精简且巧妙。
下面就一个小例子,针对不加锁、乐观锁以及悲观锁这三种方式来实现。
主要是一个用户表,它有一个年龄的字段,然后并发地对其加一,看看结果是否正确。
一些基础实现类
@Data
@Builder
@AllArgsConstructor
@NoArgsConstructor
public class User {
private Integer age;
private String name;
private Long id;
private Long version;
}
public interface UserMapper {
@Results(value = {
@Result(property = "id", column = "id", javaType = Long.class, jdbcType = JdbcType.BIGINT),
@Result(property = "age", column = "age", javaType = Integer.class, jdbcType = JdbcType.INTEGER),
@Result(property = "name", column = "name", javaType = String.class, jdbcType = JdbcType.VARCHAR),
@Result(property = "version", column = "version", javaType = Long.class, jdbcType = JdbcType.BIGINT),
})
@Select("SELECT id, age, name, version FROM user WHERE id = #{id}")
User getUser(Long id);
@Update("UPDATE user SET age = #{age}, version=version+1 WHERE id = #{id} AND version = #{version}")
Boolean compareAndSetAgeById(Long id, Long version, Integer age);
@Update("UPDATE user SET age = #{age} WHERE id = #{id}")
Boolean setAgeById(Long id, Integer age);
@Select("SELECT id, age, name, version FROM user WHERE id = #{id} for update")
User getUserForUpdate(Long id);
}
private static void exe(CountDownLatch countDownLatch, int threads, Runnable runnable) {
ExecutorService executorService = Executors.newCachedThreadPool();
for (int i = 0; i < threads; i++) {
executorService.execute(() -> {
runnable.run();
countDownLatch.countDown();
});
}
executorService.shutdown();
}
private static User getUser(long id) {
try (SqlSession session = openSession(getDataSource1())) {
UserMapper userMapper = session.getMapper(UserMapper.class);
return userMapper.getUser(id);
}
}
public static SqlSession openSession(DataSource dataSource) {
TransactionFactory transactionFactory = new JdbcTransactionFactory();
Environment environment = new Environment("development", transactionFactory, dataSource);
Configuration configuration = new Configuration(environment);
configuration.addMapper(UserMapper.class);
configuration.setCacheEnabled(false);
SqlSessionFactory sqlSessionFactory = new SqlSessionFactoryBuilder().build(configuration);
return sqlSessionFactory.openSession();
}
private static DataSource getDataSource1() {
MysqlConnectionPoolDataSource dataSource = new MysqlConnectionPoolDataSource();
dataSource.setUser("root");
dataSource.setPassword("root");
dataSource.setUrl("jdbc:mysql://localhost:3306/test?useUnicode=true&characterEncoding=utf-8");
return dataSource;
}
private static Boolean addAge(long id, int value) {
Boolean result;
try (SqlSession session = openSession(getDataSource1())) {
UserMapper userMapper = session.getMapper(UserMapper.class);
User user = userMapper.getUser(id);
result = userMapper.setAgeById(user.getId(), user.getAge() + value);
session.commit();
}
return result;
}
public static void main(String[] args) throws Exception {
long id = 1L;
int threads = 50;
CountDownLatch countDownLatch = new CountDownLatch(threads);
log.info("user:{}", getUser(id));
long start = System.currentTimeMillis();
exe(countDownLatch, threads, () -> addAge(1, 1));
countDownLatch.await();
long end = System.currentTimeMillis();
log.info("end - start : {}", end - start);
log.info("user:{}", getUser(id));
}
865 [main] INFO cn.eagle.li.mybatis.UpdateMain - user:User(age=0, name=3, id=1, version=0)
1033 [main] INFO cn.eagle.li.mybatis.UpdateMain - end - start : 164
1046 [main] INFO cn.eagle.li.mybatis.UpdateMain - user:User(age=9, name=3, id=1, version=0)
从输出可以看出,50个并发,但是执行成功的只有9个,这种实现很明显是有问题的。
private static Boolean compareAndAddAge(long id, int value, int times) {
int time = 0;
Boolean result = false;
while (time++ < times && BooleanUtils.isFalse(result)) {
result = compareAndAddAge(id, value);
}
return result;
}
private static Boolean compareAndAddAge(long id, int value) {
try (SqlSession session = openSession(getDataSource1())) {
UserMapper userMapper = session.getMapper(UserMapper.class);
User user = userMapper.getUser(id);
Boolean result = userMapper.compareAndSetAgeById(id, user.getVersion(), user.getAge() + value);
session.commit();
return result;
}
}
public static void main(String[] args) throws Exception {
long id = 1L;
int threads = 50;
CountDownLatch countDownLatch = new CountDownLatch(threads);
log.info("user:{}", getUser(id));
long start = System.currentTimeMillis();
exe(countDownLatch, threads, () -> addAge(1, 1, 20));
countDownLatch.await();
long end = System.currentTimeMillis();
log.info("end - start : {}", end - start);
log.info("user:{}", getUser(id));
}
758 [main] INFO cn.eagle.li.mybatis.UpdateMain - user:User(age=0, name=3, id=1, version=0)
1270 [main] INFO cn.eagle.li.mybatis.UpdateMain - end - start : 509
1277 [main] INFO cn.eagle.li.mybatis.UpdateMain - user:User(age=50, name=3, id=1, version=50)
从输出可以看出,并发的情况下,结果是没问题的。
private static Boolean addAgeForUpdate(long id, int value) {
Boolean result;
try (SqlSession session = openSession(getDataSource1())) {
UserMapper userMapper = session.getMapper(UserMapper.class);
User user = userMapper.getUserForUpdate(id);
result = userMapper.setAgeById(id, user.getAge() + value);
session.commit();
}
return result;
}
public static void main(String[] args) throws Exception {
long id = 1L;
int threads = 50;
CountDownLatch countDownLatch = new CountDownLatch(threads);
log.info("user:{}", getUser(id));
long start = System.currentTimeMillis();
exe(countDownLatch, threads, () -> addAgeForUpdate(1, 1));
countDownLatch.await();
long end = System.currentTimeMillis();
log.info("end - start : {}", end - start);
log.info("user:{}", getUser(id));
}
631 [main] INFO cn.eagle.li.mybatis.UpdateMain - user:User(age=0, name=3, id=1, version=50)
829 [main] INFO cn.eagle.li.mybatis.UpdateMain - end - start : 196
837 [main] INFO cn.eagle.li.mybatis.UpdateMain - user:User(age=50, name=3, id=1, version=50)
从输出可以看出,并发的情况下,结果是没问题的。
从以上来看,乐观锁和悲观锁实现都是没有问题的,至于选哪一种,还是要看业务的场景,比如说并发量的多少,加锁时长等等。
Recommend
About Joyk
Aggregate valuable and interesting links.
Joyk means Joy of geeK