MySQL transaction review

1.1 Overview of the transaction

Transaction (transaction), a transaction is a complete business logic unit, can not be divided,

For example: bank account transfer, transfer from account A to account B, you need to execute two update statements,

An update increases,

An update is reduced,

The above two update and sentences succeed and fail at the same time.

It is better to allow one success and one failure,

To ensure that the above two DML statements succeed and fail at the same time, then you need to use the database "transaction mechanism"

The statements related to transactions are only DML language, database manipulation statements, (insert delete update)

Why? Because these statements are related to the "data" in the database,

The existence of transactions is to ensure the integrity and security of data

If all businesses only need a DML language, do they still need a transaction mechanism?

No need for affairs;

But the actual situation is not like this, usually a transaction, "business", requires multiple DML languages to jointly complete

1.2 Principle of transaction

mysql five major statements

DQL database query statement
DML database manipulation statement
DDL database definition language
TCL (transaction: commit rollback) savepoint savepoint
DCL: (authorization)

1.3 The characteristics of the transaction

Four characteristics of transaction ACID

A. ( Atomicity ) Atomicity : A transaction is the smallest unit and cannot be divided
C ( consistency ) consistency: The transaction must ensure that multiple DML statements succeed or fail at the same time.
I ( isolation ) isolation: transaction A must have isolation from transaction B equivalent to thread
D ( durability ): Durability: Durability means that the final data must be persisted to the hard disk file before the transaction is considered successful.

1.4 transaction isolation

About isolation between transactions

There are isolation levels for transaction isolation, which theoretically include four:

In fact, they start in second gear:

The first level: Read uncommitted (read uncommitted), the current transaction can read the other party's uncommitted transaction

Reading uncommitted data will have dirty reads (dirty reads are reading dirty data that was not committed by the previous transaction) phenomenon: it means that dirty data has been read

The second level: **Read Committed** We can read the data submitted by the other party's transaction

The problem with read committed is: non-repeatable read (non-repeatable read is to re-read the submitted data of the previous transaction)

The third level: Repeatable Reads (repeatable read)

This isolation level solves: non-repeatable read

There is a problem: the data read is a phantom

Phantom read (virtual read) refers to reading data inserted by another transaction in one transaction, which results in inconsistent reading before and after.

The fourth level (Serializable): serialization, serialized reading solves all problems,

Disadvantages: low efficiency and need to queue up transactions

Oracle starts in second gear by default, read has been submitted

Mysql default isolation level, repeatable reading

1.5 Demo business

Mysql transaction is automatically submitted by default, (what is automatic submission, as long as any DML statement is executed, it will be submitted once)

First, set the isolation transaction level

1. read uncommitted (read uncommitted)

Read uncommitted

Open two windows, modify the data of one transaction, one transaction can be queried,

Set the isolation level of the transaction

set global transaction isolation level read uncommitted;

View msyql isolation level;

select @@global.tx_isolation;

After setting the isolation level, leave.

2. read committed (read committed)

Can read the data submitted by other transactions (the default isolation level of most databases)

3. repeatable read (repeatable read)

Mysql default isolation level

When we set the isolation level of the current conversation to repeatable read, the current conversation can be read repeatedly, that is, the result set of each read is the same. In fact, a copy of the data is backed up, and the phenomenon of phantom reading occurs.