Read mysql45 lecture - index

Index is to improve the efficiency of query

The index I understand is to store data in a certain way , In this way, the desired data can be queried faster according to the characteristics of the storage mode ,
This special storage method is the data structure , Common data structures are arrays , Linked list , Hashtable , Trees , queue , Stacks and so on .

Hashtable

Hash table is a kind of hash table with key value （key-value） Structure of stored data . for example key=monday,value= Monday ; First, according to the custom hash function monday A value is calculated 128, Then in the array array Subscript is 128 On Monday ; But maybe the hash function will calculate different values to get the same number , At this time, multiple values in the same position form a linked list .
The values calculated by the hash function here are usually not sequential values , The advantage is that it can be calculated directly when adding key Then put it in the array , It is more suitable for equivalent query ; But if you need to query a range of values (key=monday,key=friday), We need to key=monday To key=friday All the values of are queried .

Ordered array

If you use an ordered array to store data , Convenient interval query , Equivalent query is also very convenient . But it takes more space and time to insert and add . So an ordered array is suitable for storing data that normally does not change , For example, last year, Wanglaoji sold a lot in the snack bar .

The core of database underlying storage is based on data model , That is, the actual use of data structures .

InnoDB The index of

stay InnoDB in , Tables are stored by index in primary key order ,InnoDB Used B+ Tree index model , So the data is stored in B+ In the tree .
Create a table ：

 create table T( id int primary key, k int not null, name varchar(16), index (k))engine=InnoDB;

InnoDB The index types of are divided into primary key index and non primary key index .
You can see that the value stored in the leaf node of the primary key index on the left is the entire row of data (R1~R5), Primary keys are also called clustered indexes
On the right, the non primary key index leaf node stores ID Value , Non primary key indexes are also called secondary indexes .

The difference between query based on primary key index and general index ：

• select * from table where id = 5 Primary key query , Search the index of the primary key directly , That's the primary key id Composed of B+ Trees
• select * from table where k=6 General index query , First, query k The values on this field B+ Trees , find id, And then go back to according to id Composed of B+ The tree queries the entire row of values .

That is, one more step , This process is called back to table , Therefore, primary key queries should be used as much as possible .

B+ Tree to maintain the order of the index , Therefore, there will be maintenance operations when inserting data , For example, in the picture above , Insert a id The value is 400 The data of , You need to id=500,id=600 The position of the data is moved back .

During normal table creation, the primary key is set to auto increment primary key ,( namely ： NOTNULL PRIMARY KEY AUTO_INCREMENT.), So when inserting new data , There is no need to specify the primary key value , The database will automatically obtain the current maximum primary key value, and then +1 Assign to new data . However, the primary keys of some tables are not self incremented , Instead, use fields with business logic as primary keys .
Consider whether to use the ID number as the primary key of the personnel table or the auto increment primary key ？
If you use the ID number as the primary key , Then the value stored in the secondary index needs 20 Bytes , But if you use a self incrementing primary key , The value stored in the secondary index is 4 byte (int),8 byte (bigint), You can see that the space occupied by the self incrementing primary key index is still a little smaller .

It is suitable for the background of using business fields as primary keys ：

• Only one index
• The index is the only index

In this case, there is no need to consider the secondary index . But normally, it is convenient to add a primary key automatically .