Home >Database >Mysql Tutorial >In-depth understanding of transactions, 4 major features, and isolation levels in MySQL
This article is an advanced study of MySQL. It will give you a detailed introduction to transactions in MySQL, the four major features (ACID) and the isolation level of transactions. I hope it will be helpful to you!
[Related recommendations: mysql video tutorial]
The environment version used for the operation and testing of this article is 5.7.21
mysql> select version(); +-----------+ | version() | +-----------+ | 5.7.21 | +-----------+ 1 row in set (0.00 sec)
Remember: Only InnoDB among our common MySQL storage engines supports transactions. Therefore, the following operations are also done under InnoDB.
1. What is a transaction
A more classic case is bank transfer: Little A transfers 100 yuan to Little B
Normal situation: Little A’s account is deducted by 100 yuan, and Little B’s account is increased by 100 yuan.
Abnormal situation: Little A’s account is deducted by 100 yuan, and the amount in Little B’s account remains unchanged.
Under abnormal circumstances, there was a problem with the banking system after 100 yuan was deducted from Little A's account, and the operation of adding 100 yuan to Little B's account was not executed. That is to say, the amounts on both sides do not match up. Little A is not willing, Little B is not willing, and neither is the bank. The purpose of affairs is to avoid abnormal situations and satisfy everyone.
2. The 4 major characteristics of transactions (ACID)
The operations of the transaction are indivisible, either all operations or none, just like transfers, there is no intermediate state. And this atomicity does not mean that there is only one action, there may be many operations, but it is indivisible from the result, which means that atomicity is a result state.
Before and after executing the transaction, the data remains consistent, just like the bank account system. Regardless of whether the transaction is successful or not, the total account amount of the two It should be the same.
When multiple transactions operate data at the same time, multiple transactions are directly isolated from each other and will not affect each other.
The impact of a transaction on data after submission is permanent and will not be lost when written to disk.
3. Explicit transactions and implicit transactions
mysql transactions are divided into explicit transactions
andImplicit transaction
, the default transaction is an implicit transaction, which is automatically opened, committed, and rolled back during operation by the variable autocommit
.
The key commands for control are as follows
set autocommit=0; -- 关闭自动提交事务(显式) set autocommit=1; -- 开启自动提交事务(隐式) -- 当autocommit=0的时候手动控制事务 rollback; -- 回滚事务 commit; -- 提交事务 -- 当autocommit=1 自动提交事务,但是可以控制手动提交 start transaction; -- 开启事务(或者用begin开启事务) commit; -- 提交事务 rollback; -- 回滚事务 SAVEPOINT 保存点名称; -- 保存点(相当于存档,可以不用回滚全部操作) rollback to 保存点; -- 回滚到某个保存点 (这个后面就不测试,知道有这个操作就行)
First create a table ajisun
mysql> create table ajisun(id int(5), name varchar(20) character set utf8 COLLATE utf8_bin ) engine=innodb character set= utf8mb4 COLLATE = utf8mb4_bin; Query OK, 0 rows affected (0.03 sec)
1. Implicit transaction
-- 看下当前autocommit的状态是,默认是on状态 mysql> show variables like 'autocommit'; +---------------+-------+ | Variable_name | Value | +---------------+-------+ | autocommit | ON | +---------------+-------+ 1 row in set (0.01 sec) -- 插入一条数据 mysql> insert into ajisun values(1,'阿纪'); Query OK, 1 row affected (0.00 sec) mysql> rollback; -- 执行rollback 也是没有效果的,还是能够查询到插入的数据(不需要我们手动控制commit) mysql> select * from ajisun; +------+--------+ | id | name | +------+--------+ | 1 | 阿纪 | +------+--------+ 1 row in set (0.00 sec)
2. Explicit transaction mode 1
Explicit transactions allow us to control the opening, submission, rollback and other operations of the transaction
-- 开启显式事务-回滚 mysql> set autocommit=0; Query OK, 0 rows affected (0.00 sec) mysql> select * from ajisun; +------+--------+ | id | name | +------+--------+ | 1 | 阿纪 | +------+--------+ 1 row in set (0.00 sec) mysql> insert into ajisun values(2,'纪先生'); Query OK, 1 row affected (0.00 sec) -- 插入后可以看见2条数据 mysql> select * from ajisun; +------+-----------+ | id | name | +------+-----------+ | 1 | 阿纪 | | 2 | 纪先生 | +------+-----------+ 2 rows in set (0.00 sec) -- 回滚之后上面插入的数据就没了 mysql> rollback; Query OK, 0 rows affected (0.00 sec) mysql> select * from ajisun; +------+--------+ | id | name | +------+--------+ | 1 | 阿纪 | +------+--------+ 1 row in set (0.00 sec)
-- 插入一条数据 mysql> insert into ajisun values(2,'ajisun'); Query OK, 1 row affected (0.01 sec) -- 提交 mysql> commit; Query OK, 0 rows affected (0.00 sec) -- 回滚 mysql> rollback; Query OK, 0 rows affected (0.00 sec) -- 先提交commit,在rollback 数据依然存在,说明commit生效,事务已提交,回滚就不生效了。 mysql> select * from ajisun; +------+--------+ | id | name | +------+--------+ | 1 | 阿纪 | | 2 | ajisun | +------+--------+ 2 rows in set (0.00 sec)
3. Explicit transactions Method 2
Use start transaction
Change to the default transaction firstset autocommit=1;
-- 开启事务 mysql> start transaction; Query OK, 0 rows affected (0.00 sec) mysql> delete from ajisun where id=1; Query OK, 1 row affected (0.00 sec) -- 提交事务 mysql> commit; Query OK, 0 rows affected (0.01 sec) mysql> select * from ajisun; +------+--------+ | id | name | +------+--------+ | 2 | ajisun | +------+--------+ 1 row in set (0.00 sec)
-- 开启事务 mysql> start transaction; Query OK, 0 rows affected (0.00 sec) mysql> delete from ajisun where id =2; Query OK, 1 row affected (0.01 sec) -- 回滚事务 mysql> rollback; Query OK, 0 rows affected (0.01 sec) -- 删除操作失效了 mysql> select * from ajisun; +------+--------+ | id | name | +------+--------+ | 2 | ajisun | +------+--------+ 1 row in set (0.00 sec)
4. Problems in concurrent transactions
If there is only one transaction operating on the table at the same time, there will be no problem, but this is impossible. In reality, it is used as much as possible, and multiple transactions are operated at the same time. Multiple transactions will bring many problems, such as Dirty read
, Dirty write
, `Non-repeatable read
, phantom read
1. Dirty read
#A transaction reads modified data from another uncommitted transaction. This is a dirty read.
For example, two transactions a and b: operate a record at the same time
After transaction a modifies the record, it has not been officially submitted to the database. At this time, transaction b reads it, and then uses read The obtained data is used for subsequent operations.
If transaction a is rolled back and the modified data no longer exists, then transaction b is using non-existent data. This is dirty data.
2. Dirty write (data loss)
One transaction modified the data modified by another uncommitted transaction
For example, two transactions a and b: operate a record at the same time
A transaction is not submitted after modification, then b transaction also modifies the same data, and then b transaction commits the data.
If transaction a rolls back its own modifications and also rolls back the modifications of transaction b, the problem will be: transaction b has been modified and submitted, but the database has not changed. This situation is Dirty writing.
3. Non-repeatable read
A transaction can only read the data modified by another committed transaction, and other transactions cannot read the data. After a modification is made and submitted, the transaction can be queried to get the latest value.
也就是在同一个事务中多次读取同一条记录,得到的内容都不一样(在每次读取之前都有其他事务完成修改并提交),这就是不可重复读
。
4. 幻读
在一个事务内 相同条件查询数据,先后查询到的记录数不一样
也就是一个事务先根据某些条件查询出一些记录,之后另一个事务又向表中插入了符合这些条件的记录,原先的事务再次按照该条件查询时,能把另一个事务插入的记录也读出来,那就意味着发生了幻读
不可重复读和幻读的区别:不可重复读重点在于同一条记录前后数据值不一样(内容的变化),而幻读重点在于相同查询条件前后所获取的记录数不一样(条数的变化)
五. 事务的隔离级别
上面说的事务的并发问题,在不同的场景下要求不一样,能接受的问题也不一样。他们之间的严重性排序如下:
脏写 > 脏读 > 不可重复读 > 幻读
MySQL中提供了4种隔离级别来处理这几个问题,如下
隔离级别 | 脏读 | 不可重复读 | 幻影读 |
---|---|---|---|
READ- UNCOMMITTED | √ | √ | √ |
READ-COMMITTED | × | √ | √ |
REPEATABLE-READ | × | × | √ |
SERIALIZABLE | × | × | × |
SQL 标准定义了四个隔离级别:
脏读
、不可重复读
和幻读
。但是并发度最高幻读
和不可重复读
仍有可能发生。脏读
和不可重复读
,但幻读
仍有可能发生。脏读
、不可重复读
以及幻读
。并发度也是最低的MySQL默认采用的 REPEATABLE_READ 隔离级别 Oracle默认采用的 READ_COMMITTED 隔离级别
1. 如何设置隔离级别
可以通过变量参数transaction_isolation
查看隔离级别
mysql> SELECT @@transaction_isolation; +-------------------------+ | @@transaction_isolation | +-------------------------+ | REPEATABLE-READ | +-------------------------+ 1 row in set (0.00 sec) mysql> show variables like '%transaction_isolation%'; +-----------------------+-----------------+ | Variable_name | Value | +-----------------------+-----------------+ | transaction_isolation | REPEATABLE-READ | +-----------------------+-----------------+ 1 row in set (0.02 sec)
修改的命令:SET [GLOBAL|SESSION] TRANSACTION ISOLATION LEVEL $[level];
level的值就是4中隔离级别READ-UNCOMMITTED
READ-COMMITTED
REPEATABLE-READ
SERIALIZABLE
设置全局隔离级别
只对执行完该语句之后产生的会话起作用。
当前已经存在的会话无效。
set global transaction_isolation='read-uncommitted'; set global transaction_isolation='read-committed'; set global transaction_isolation='repeatable-read'; set global transaction_isolation='serializable';
例如:
会话A
mysql> set global transaction_isolation='serializable'; Query OK, 0 rows affected (0.01 sec) mysql> select @@global.transaction_isolation; +--------------------------------+ | @@global.transaction_isolation | +--------------------------------+ | SERIALIZABLE | +--------------------------------+ 1 row in set (0.00 sec) -- 当前会话(设置之前就已经存在的会,级别是默认的) mysql> select @@transaction_isolation; +-------------------------+ | @@transaction_isolation | +-------------------------+ | REPEATABLE-READ | +-------------------------+ 1 row in set (0.00 sec)
会话B(set之后新建的会话)
mysql> select @@global.transaction_isolation; +--------------------------------+ | @@global.transaction_isolation | +--------------------------------+ | SERIALIZABLE | +--------------------------------+ 1 row in set (0.00 sec) mysql> select @@transaction_isolation; +-------------------------+ | @@transaction_isolation | +-------------------------+ | SERIALIZABLE | +-------------------------+ 1 row in set (0.00 sec)
设置会话的隔离级别
对当前会话的所有后续的事务有效
该语句可以在已经开启的事务中间执行,但不会影响当前正在执行的事务。
如果在事务之间执行,则对后续的事务有效。
set session transaction_isolation='read-uncommitted'; set session transaction_isolation='read-committed'; set session transaction_isolation='repeatable-read'; set session transaction_isolation='serializable';
比如:
会话A
mysql> set session transaction_isolation='read-uncommitted'; Query OK, 0 rows affected (0.00 sec) mysql> select @@transaction_isolation; +-------------------------+ | @@transaction_isolation | +-------------------------+ | READ-UNCOMMITTED | +-------------------------+ 1 row in set (0.00 sec)
新建会话B(依然是默认的级别:可重复读)
mysql> select @@transaction_isolation; +-------------------------+ | @@transaction_isolation | +-------------------------+ | REPEATABLE-READ | +-------------------------+ 1 row in set (0.00 sec)
2. 怎么选择隔离级别
一般情况下默认的可重复读
就好了,一般很少改这个,除非业务场景特殊
记住一点:隔离级别越高,并发问题就越少,但并发性也就越低,所以还是要根据业务选择来。
六. 总结
事务的四大特性:原子性,一致性,隔离性,持久性
事务的常见命令:
set autocommit=0/1; -- 关闭/开启自动提交事务 start transaction; -- 开启事务(或者用begin) rollback; -- 回滚事务 commit; -- 提交事务
并发事务的问题:脏写 > 脏读 > 不可重复读 > 幻读
需要熟悉事务的4种隔离级别以及MySQL默认级别
怎么设置隔离级别(global,session)
更多编程相关知识,请访问:编程入门!!
The above is the detailed content of In-depth understanding of transactions, 4 major features, and isolation levels in MySQL. For more information, please follow other related articles on the PHP Chinese website!