Interview questions MySQL database

Multi table associated query

Relational query sql The idea of writing ：1. First determine the connected table ;2. Determine the field to query ;3. Determine the connection conditions and connection methods .

topic

Suppose there is t_employee( The employee table ) and t_dept( Departmental table ) For example ：
t_employee In the table dept On behalf of the employee's department

1. Query the employee name and its corresponding department name （ There is no department , If there is no employee in the Department, it will not be displayed ）
2. Check the names of all employees and their departments （ No department is displayed as null）
3. Query all department names and their corresponding employee information （ No employee displayed null）
4. Query the information of all employees and all departments （ No departments or employees are displayed null）
5. Check the names of employees and their superiors

1. Internal connection query

SELECT e.emp_name, d.dept_name FROM t_employee e INNER JOIN t_dept d ON e.dept=d.id
SELECT e.emp_name, d.dept_name FROM t_employee e, t_dept d WHERE e.dept=d.id

2. Left outer connection query

SELECT e.emp_name, d.dept_name FROM t_employee e LEFT JOIN t_dept d ON e.dept=d.id

3. Right outer connection query

SELECT d.dept_name, e.* FROM t_employee e RIGHT JOIN t_dept d ON e.dept=d.id

4. All external connection query

SELECT e.*, d.* FROM t_employee e LEFT JOIN t_dept d ON e.dept=d.id
UNION
SELECT e.*, d.* FROM t_employee e RIGHT JOIN t_dept d ON e.dept=d.id
ps: stay oracle in , Use... Directly full outer join Just connect two tables with keywords 

5. Self connect query

Self join query is the join query between the current table and itself , The key point is to virtualize a table to an alias .
SELECT e.emp_name, d.emp_name FROM t_employee e LEFT JOIN t_employee d ON e.boss_id = d.id

What is the index ？

1. Greatly reduce the amount of scanning required by the server .
2. Help server avoid sorting and temporary tables
3. Random IO Into order IO, Reduce track addressing time

Index type ？ Primary key 、 only 、 Ordinary 、 The full text 、 Composite index

B+Tree

B+Tree yes B-Tree Extension of contract
Underlying data structure

There are several indexes in a table B+ Trees .

How many indexes can there be in a table ？

Look at the execution plan .id key,keylength Extral
key Represents whether the index will be used in the current query .
type： Access type .const,system,ref,range,all

Several copies of actual data are stored in a table ？

When inserting data in a table , Must be bound to an index to store .
The index is selected in the order of having a primary key and using a primary key , There is no primary key but a unique key . If there is no unique key, use 6 Bytes of rowid.
Other indexes B+ The leaf node in the tree structure is the data binding index key value .

Cluster index 、 Nonclustered index

Cluster index ： Index stored with data binding , Such as id.
Nonclustered index ： An index stored separately from data , Such as name.
InnoDB Is it clustered or non clustered
InnoDB There is at least one clustered index , There can be multiple non clustered indexes .
If there is no index , The system generates a row_id.
.frm Table structure ,.ibd Data in the + Index file .

2.MyISAM Are non clustered indexes ..frm .myd .myi

Why use int type id Self increasing good

because InnoDB The default primary key of the engine is cluster index , The physical storage order of the cluster index data is consistent with the index order , As long as the index is adjacent , Then the data is also on the adjacent disks .（int type id Self increasing , This is related to the data structure of the database index .）

Index optimization

1 Back to the table

The index value is found according to the normal column , Go back to the primary key to find .

6 Bytes of role_id

2 Index overlay （select *）

select id, name FROM Student WHERE id=1（ There is no need to return the form ）
select * FROM Student WHERE id=1（ We need to go back to the table ）

3 The leftmost match

The optimizer automatically matches
Enforcement

4 Index push down

Processing in the storage engine

OR Can you walk the index ？

answer ： The primary key will .

Optimize the details

1. When using indexed columns , Try not to use expressions .
2. Try to use primary keys , Self increasing （ The leaves split ） It doesn't go back to the table .
3. Put a string before x Bit create index .
4. remarks 、 Or the Chinese name field should not be indexed . For example, the discrimination of remark information is not high , Do not add index .
5. Use index scan to sort .（ Cost performance ）
6. topic ： use 256M Memory handle 1T File sorting .key_Length Calculation method .
7. where In conditional query , Try not to use intermediate parameters > or <
8. Cast , Full table scan .（ Integer to string conversion , Implicit conversion ）
9. Total number of rows is greater than 80%, Need to index . If it is less than, there is no need to build an index .
10. The columns that create the index , Not allowed to be empty .Mybatis Pit handle 0 When it comes to null.
11. When the meter is connected , Not more than 3 A watch .（ Minimize data redundancy ）
12. limiit
13. Number of single table indexes 5 Within a .
14. The number of multi column single index fields shall not exceed 5 individual .
15. myth ：① The more indexes, the better ;② Premature optimization

Explain Implementation plan

topic ： use 256M Memory handle 1T File sorting ？

Reference address

once MySQL Index interview
https://mp.weixin.qq.com/s/65V-htzeAU1ACBvga8MlaQ
What is the pessimistic lock and the optimistic lock ？
https://mp.weixin.qq.com/s/I5hHbSeJZlpGcRhV1vOBnw

Why index can improve query speed ？
https://mp.weixin.qq.com/s/BTCZikV60bZPWCtZ1to6Qg