MySQL中的行级锁定

654 阅读3分钟

锁是在执行多线程时用于强行限定资源访问的同步机制,数据库锁根据锁的粒度可分为行级锁,表级锁和页级锁

行级锁

行级锁是mysql中粒度最细的一种锁机制,表示只对当前所操作的行进行加锁,行级锁发生冲突的概率很低,其粒度最小,但是加锁的代价最大。行级锁分为共享锁和排他锁。

特点:

开销大,加锁慢,会出现死锁;锁定粒度最小,发生锁冲突的概率最大,并发性也高;

实现原理:

InnoDB行锁是通过给索引项加锁来实现的,这一点mysql和oracle不同,后者是通过在数据库中对相应的数据行加锁来实现的,InnoDB这种行级锁决定,只有通过索引条件来检索数据,才能使用行级锁,否则,直接使用表级锁。特别注意:使用行级锁一定要使用索引

举个栗子:

创建表结构

CREATE TABLE `developerinfo` (
  `userID` bigint(20) NOT NULL,
  `name` varchar(255) DEFAULT NULL,
  `passWord` varchar(255) DEFAULT NULL,
  PRIMARY KEY (`userID`),
  KEY `PASSWORD_INDEX` (`passWord`) USING BTREE
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

插入数据

INSERT INTO `developerinfo` VALUES ('1', 'liujie', '123456');
INSERT INTO `developerinfo` VALUES ('2', 'yitong', '123');
INSERT INTO `developerinfo` VALUES ('3', 'tong', '123456');
(1)通过主键索引来查询数据库使用行锁

打开三个命令行窗口进行测试

命令行窗口1 命令行窗口2 命令行窗口3

mysql> set autocommit = 0; Query OK, 0 rows affected mysql> select * from developerinfo where userid = '1' for update; +--------+--------+----------+ | userID | name | passWord | +--------+--------+----------+ | 1 | liujie | 123456 | +--------+--------+----------+ 1 row in set |mysql> set autocommit = 0; Query OK, 0 rows affected mysql> select * from developerinfo where userid = '1' for update; 等待

|mysql> set autocommit = 0; Query OK, 0 rows affected mysql> select * from developerinfo where userid = '3' for update; +--------+------+----------+ | userID | name | passWord | +--------+------+----------+ | 3 | tong | 123456 | +--------+------+----------+ 1 row in set |mysql> commit; Query OK, 0 rows affected mysql> select * from developerinfo where userid = '1' for update; +--------+--------+----------+ | userID | name | passWord | +--------+--------+----------+ | 1 | liujie | 123456 | +--------+--------+----------+ 1 row in set

(2)查询非索引的字段来查询数据库使用行锁

打开两个命令行窗口进行测试

命令行窗口1 命令行窗口2

|mysql> set autocommit=0; Query OK, 0 rows affected mysql> select * from developerinfo where name = 'liujie' for update; +--------+--------+----------+ userID name passWord
+--------+--------+----------+ 1 liujie 123456
+--------+--------+----------+ 1 row in set |mysql> set autocommit=0; Query OK, 0 rows affected mysql> select * from developerinfo where name = 'tong' for update; 等待|

mysql> commit; Query OK, 0 rows affected mysql> select * from developerinfo where name = 'liujie' for update; +--------+--------+----------+ | userID | name | passWord | +--------+--------+----------+ | 1 | liujie | 123456 | +--------+--------+----------+ 1 row in set

##### (3)查询非唯一索引字段来查询数据库使用行锁锁住多行

mysql的行锁是针对索引假的锁,不是针对记录,所以可能会出现锁住不同记录的场景

打开三个命令行窗口进行测试

命令行窗口1	命令行窗口2	命令行窗口3

mysql> set autocommit=0; Query OK, 0 rows affected mysql> select * from developerinfo where password = '123456 ' for update; +--------+--------+----------+ | userID | name | passWord | +--------+--------+----------+ | 1 | liujie | 123456 | | 3 | tong | 123456 | +--------+--------+----------+ 2 rows in set mysql> set autocommit =0 ; Query OK, 0 rows affected mysql> select * from developerinfo where userid = '1' for update;


等待

mysql> set autocommit = 0; Query OK, 0 rows affected mysql> select * from developerinfo where userid = '2 ' for update; +--------+--------+----------+ | userID | name | passWord | +--------+--------+----------+ | 2 | yitong | 123 | +--------+--------+----------+ 1 row in set commit; mysql> select * from developerinfo where userid = '1' for update; +--------+--------+----------+ | userID | name | passWord | +--------+--------+----------+ | 1 | liujie | 123456 | +--------+--------+----------+ 1 row in set

##### (4)条件中使用索引来操作检索数据库时,是否使用索引还需有mysql通过判断不同执行计划来决定,是否使用该索引,如需判定如何使用explain来判断索引,请听下回分解