MySQL four locks

Row-level locks ：

• stay mysql in , Row level lock is not a direct lock record , It's the lock index , Index is divided into primary key index and non primary key index .
• One SQL Statement operates the primary key index , that MySQL Lock the primary key index .
• One SQL Statement operates on a non primary key index , that MySQL Lock the non primary key index first , Then lock the relevant non primary key index .
• stay InnoDB in , If SQL The statement does not involve an index , Then the records will be locked through the hidden cluster index . That is, the data is not searched through the index , that InnDB All the data in the table will be locked . No index , When searching records, you need to scan the whole table , Scan the whole table , Then lock the table .
• row-level loking , Lock a row of records
• Spending big , Full lock , There will be a deadlock .
• Small particle size of cable , The probability of lock conflict is small , High concurrency efficiency .
• Suitable for concurrent writing .

Table lock ：

 

• Table lock , Generally, it refers to table structure shared lock , Yes, this table cannot be executed DDL operation , But yes DML Operation is not limited . Before row level locks, you need to add table structure shared locks . If you use a table lock , Other customers will not be able to query , Therefore, row level locks are generally used in development .
• table-level locking, Lock the whole table .
• Low overhead , Locked fast , No deadlock .
• Big lock size , The probability of lock conflict is high , Concurrency is inefficient .
• Suitable for inquiry .

Optimism lock ：

• Optimistic locking believes that concurrent operations are difficult to occur , But it doesn't necessarily happen , All measures exist to deal with this situation , That's version number version, Every time you modify the data ,version Change once .
• for example ,A The window is set on the day of 12:00 The stage operation updates the data in the table ,12:02 Press the OK key . But on the same day 12:01 Stage B The window updates and determines the data in the table , Then the version number of the table version It has changed , At the same time A Data update of window operation failed . This also indicates that concurrency problems have occurred , This operation needs to be rolled back .
• When reading data , No locks .

Pessimistic locking ：

• And optimism , Pessimistic locks believe that concurrency problems are very easy to occur .
• A When the window is updating data ,B The window cannot manipulate this data ,B The window wants to update data , Must wait A Only after the window operation is completed .
• During data operation , Whether reading or writing , Will lock the records , To prevent other threads from changing data .