MySQL summary

1、mysql What are the indexes
     primary key He has to be primary key Each table can have only one index
     A normal index is also called a secondary index , Accelerated search
     unique index ：
              primary key ：primary key ： Accelerated search + constraint （ Not empty and unique ）
             unique index ：unique： Accelerated search + constraint （ only ）
     Joint index ：
         Follow the prefix principle
     Full-text index ：
      When searching for a long article , The best effect .

2、 Clustered index and non clustered index
     The difference between a clustered index and a nonclustered index is ： Whether the leaf node stores a whole row of records
    InnoDB The primary key uses a clustered index ,MyISAM Whether it's a primary key index , Whether the secondary index uses non clustered index .

3、innodb and myisam The difference between
   innodb Support things ,myisam Do not support things
    Cluster index , Nonclustered index
    Row lock , Table locks
    Full text indexing is not supported , Full-text index
   InnoDB Support foreign keys , and MyISAM I won't support it
4、innodb Engine 4 Big characteristic
     Insert buffer （insert buffer), The two time to write (double write), adaptive hash index (ahi), read-ahead (read ahead)
5、MySQL The principle and process of replication
    Mainly depends on

    
6、 Characteristics of things acid
     Atomicity （Atomicity, Or indivisibility ）
     Uniformity （Consistency）
     Isolation, （Isolation）
     persistence （Durability）
    InnoDB The storage engine also provides two types of transaction logs ：redo log( Redo log ) and undo log( Rollback log ). among redo log Used to ensure transaction persistence ;undo log Is the basis for atomicity and isolation of transactions .

7、 redo log And binlog
     We know , stay MySQL There is still binlog( Binary log ) Write operations can also be recorded and used for data recovery , But they are fundamentally different ：

    （1） The effect is different ：redo log Is used for crash recovery Of , Guarantee MySQL Downtime also does not affect persistence ;binlog Is used for point-in-time recovery Of , Ensure that the server can recover data based on time points , Besides binlog Also used for master-slave replication .

    （2） Different levels ：redo log yes InnoDB Storage engine implementation , and binlog yes MySQL The server layer of ( You can refer to the front of the article MySQL Introduction to logical architecture ) Realized , Support at the same time InnoDB And other storage engines .

    （3） The content is different ：redo log It's a physical log , Content is disk-based Page;binlog The content of is binary , according to binlog_format Parameter difference , It may be based on sql sentence 、 Based on the data itself or a mixture of the two .

    （4） Different write times ：binlog Write at transaction commit time ;redo log The write time is relatively multivariate ：

     As mentioned earlier ： Called when a transaction commits fsync Yes redo log To brush plate ; This is the default policy , modify innodb_flush_log_at_trx_commit Parameters can change the policy , But the durability of the transaction will not be guaranteed .
     Except when the transaction commits , There are other brush times ： Such as master thread Swipe the disk once a second redo log etc. , The advantage is that it doesn't have to wait commit When the brush set ,commit The speed has been greatly accelerated .

8、 The transaction isolation
   1、 Read uncommitted ( Dirty reading ) Uncommitted data read
   2、 Read submitted ( It can't be read repeatedly ) What I read is the latest log,
   3、 Repeatable ( Fantasy reading ) -- resolvent . The difference between unrepeatable reading and phantom reading can be popularly understood as ： The first is that the data changed , The latter is that the number of rows of data has changed .（ adopt
   4、RR To solve the dirty read 、 It can't be read repeatedly 、 Illusory reading and so on , It uses MVCC：MVCC Full name Multi-Version Concurrency Control, Multiple versions of the concurrency control protocol .
9、mvcc
    The solution is dirty reading 、 It can't be read repeatedly 、 The problem of unreal reading
   gap and next_lock

10、B+ Trees
     The underlying leaf node is a two-way linked list

11、mysql Optimize
    configuration ：
        1、buffer_pool Try to 75%-80%
        2、 Size the number of connections
        3、 modify wait_timeout The default time for is from 8 Hours changed to 1 Hours
        4、 More open cache than write , Otherwise, it is recommended to close
    Design optimization ：
         Create table time , Available fields can be used tinyint Do not use int.
         The more indexes you create, the better , When you consider creating an index, you also need to create .
    In terms of use
        Return only the required fields .
        Do not use null, Does not occupy a space , It is no longer in the index range .
        Less use like
       

A very referential address ：

https://www.cnblogs.com/kismetv/p/10331633.html

https://juejin.im/post/6844903824935632909