Practical SQL statements and execution plan view

see SQL Implementation plan of

When the data scale is small , Queries are all done in an instant , But when the data scale increases to more than ten million, the same sql There is no output , At this time, the importance of query optimization and indexing is reflected

How to view sql Implementation plan of ：EXPLAIN select * from user where age = ‘20’;

id ：select The order in which statements are executed ,id Same from top to bottom ,id The higher the value, the higher the priority , Execute first ;

select_type：

• simple： Does not contain subqueries or union
• primary： Include subqueries , The outermost query is marked as primary
• subquery： Subquery
• union：union
• derived： Appear in the from Subquery in Clause

table： Table name

partitions： The partition table shows partitions id, The non partitioned table shows null

type：

• eq_ref ： Unique index or primary key ​
• ref： Non unique index ​
• ref_or_null： Similar to the previous query , But it's attached null Value judgement

possible_keys： The index involved in the query （ Full index ）
key： Query the index actually used , Display without index null

key_len： Maximum length of index field

ref： Indicates which columns are used to find values on indexed columns

rows： Estimate the number of rows that the lock needs to read

Economic SQL：

1、 Insert or replace

​ Insert a new record (INSERT), But if the record already exists, delete it first and then insert it

	REPLACE INTO DIS_CUS_LIST (ID,CUS_ID,CUS_NAME,BELONG_BUS,CERT_CODE) VALUES 
	(null,'0078045','xiaoxiong','02','789456123123456')

2、 Insert or update

​ Insert a new record (INSERT), But if the record already exists , Update this record

//  Use the primary key to determine whether there is 
	INSERT INTO ... ON DUPLICATE KEY UPDATE

INSERT INTO DIS_CUS_LIST (ID,CUS_ID,CUS_NAME,BELONG_BUS,CERT_CODE) VALUES 
(null,'0078045','xiaoxiong','02','789456123123456')
ON DUPLICATE KEY UPDATE BELONG_BUS='01',CERT_CODE='369852147'

3、 Insert or ignore

​ Insert a new record (INSERT), But if the record already exists , Just ignore

INSERT IGNORE INTO DIS_CUS_LIST (ID,CUS_ID,CUS_NAME,BELONG_BUS,CERT_CODE) VALUES 
(null,'0078045','xiaoxiong','02','789456123123456')

4、SQL Medium if-else Judge

​ SQL Often used in if-else Judge

CASE WHEN ... THEN ... ELSE ... END
//  scene ： Game Festival feedback ,2021 All adult males can receive... For newly registered users in 10 Yuan's red envelope , Other users can receive 5 Yuan's red envelope 
UPDATE user_info u SET u.balance = CASE WHEN u.sex = ' male ' 
and u.age > 18 THEN u.balance + 10 ELSE u.balance + 5 END WHERE u.creat_time >= '2021-01-01'

5、 Specify data snapshot or backup

​ If you want to assign data from one table to another new table , Can combine CREAT TABLE and SELECT

CREAT TABLE dis_cus_list1 SELECT * FROM dis_cus_list where belong_bus = '01'

6、 Write query result set

​ If the result set of the query needs to be written to the table , So let's create a temporary table to store data , Then use the corresponding query statement to import the result set into the table

INSERT INTO temp_1 (cus_id,belong_bus,aver_age,cert_code) 
SELECT cus_id,belong_bus,AVG(score),cert_code FROM dis_cus_list GROUP BY cus_id

7、 Force the use of the specified index

​ When we write a query statement , We know which index we need to use as the best solution , Then you can use FORCE INDEX The specified index is used for forced query

//  The specified index must exist 
SELECT * FROM dis_cus_list FORCE INDEX (idx_cus_id) WHERE cus_id = '0078045' ORDER BY id DESC

Work experience ：

​ During the work development process, there was a requirement to insert the query results into other tables , The stupidest thing is that I used the circular method to check the primary key first and then insert , As a result, if you need to insert thousands of data , You need to access the database thousands of times , Not only slow , It also increases the burden on the server ; Later, the technical manager used a query , Insert all the required data , You only need to access the database once , The efficiency is improved dozens of times in an instant .

Economic SQL Learn from ：
https://blog.csdn.net/qq_39390545/article/details/106690916