The SQL query statement executes select (1) first, and the ByteDance algorithm engineer is interviewed

SELECT

< Back to the data list > # The returned single column must be in group by clause , Except aggregate functions

DISTINCT

# Data De duplication

ORDER BY

< Sorting conditions > # Sort

LIMIT

< Row limit >

• Actually , When the engine performs each of these steps , Will form a virtual table in memory , And then do the following operation on the virtual table , And free the memory of the unused virtual table , And so on .

Specific explanation ：（ notes ： below “VT” Express → Virtual table virtual ）

1. from：**select * from table_1, table_2; And select * from table_1 join table_2; The result is the same , They all mean Cartesian product ;** It is used to directly calculate the Cartesian product of two tables , Get the virtual table VT1, This is all. select Statement is the first operation to be performed , Other operations are performed on this table , That is to say from What the operation accomplished

2. on: from VT1 Filter the data that meets the criteria in the table , formation VT2 surface ;

3. join: Will be join Type of data added to VT2 In the table , for example left join The remaining data from the left table will be added to the virtual table VT2 in , formation VT3 surface ; If the number of tables is greater than 2, It will repeat 1-3 Step ;

4. where: Perform screening ,（ You can't use aggregate functions ） obtain VT4 surface ;

5. group by: Yes VT4 Tables are grouped , obtain VT5 surface ; After processing the statement , Such as select,having, The columns used must be included in group by In the condition , Those that don't appear need aggregate functions ;

6. having: Filter grouped data , obtain VT6 surface ;

7. select: Return the column to get VT7 surface ;

8. distinct: Used to get rid of duplication VT8 surface ;

9. order by: Used to sort to get VT9 surface ;

10. limit: Returns the number of rows required , obtain VT10;

It should be noted that ：

• group by In the condition , Each column must be a valid column , It can't be an aggregate function ;

• null Values are also returned as a group ;

• Except for aggregate functions ,select Columns in clause must be in group by In the condition ;

The above shows us what a query will return , meanwhile , Also answer the following questions ：

• Can be in GRROUP BY Then use WHERE Do you ？（ no way ,GROUP BY Is in WHERE after ！）

• Can I filter the results returned by window functions ？（ no way , The window function is SELECT In the sentence , and SELECT Is in WHERE and GROUP BY after ）

• Can be based on GROUP BY What's going on in ORDER BY Do you ？（ Sure ,ORDER BY Basically at the end of the execution , So it can be based on anything ORDER BY）

• LIMIT When is the execution ？（ In the end ！）

however , The database engine does not have to be executed strictly in this order SQL Inquire about , Because in order to execute queries faster , They will make some optimizations , These questions are explained below ↓↓↓.

SQL The alias in will affect SQL Execution order ？

======================

The following parties SQL Shown ：

SELECT CONCAT(first_name, ' ', last_name) AS full_name, count(*)

FROM table

GROUP BY full_name

From this statement , As if GROUP BY Is in SELECT After that , Because it quotes SELECT One of the aliases in . But it doesn't have to be , The database engine will rewrite the query like this ↓↓↓：

SELECT CONCAT(first_name, ' ', last_name) AS full_name, count(*)

FROM table

GROUP BY CONCAT(first_name, ' ', last_name)

therefore , such GROUP BY Still execute first .

in addition , The database engine will also do a series of checks , Make sure SELECT and GROUP BY The things in are effective , So we will check the query before generating the execution plan .

The database is likely to perform queries out of the normal order （ Optimize ）

====================

In practice , The database does not necessarily follow JOIN、WHERE、GROUP BY Order to execute the query , Because they do a series of optimizations , Disorganize the order of execution , So that the query can execute faster , As long as the query results are not changed .

This query shows why queries need to be executed in different order ：

SELECT * FROM

dept d LEFT JOIN student s ON d.student_id = s.id



#  Pleasantly surprised 

 Finally, I prepared a set of interview questions corresponding to the above materials （ There's an answer ） And the high-frequency interview algorithm questions during the interview （ If the interview preparation time is not enough , Then concentrate on these algorithm problems , Hit rate 85%+）

![image.png](https://img-blog.csdnimg.cn/img_convert/a2f6f9cda63f0e1b601d8a25895dd31a.png)


![image.png](https://img-blog.csdnimg.cn/img_convert/c3838a44e11db29168828428e1106450.png)

 I also prepared a set of interview questions corresponding to the above materials （ There's an answer ） And the high-frequency interview algorithm questions during the interview （ If the interview preparation time is not enough , Then concentrate on these algorithm problems , Hit rate 85%+）

[ Outside the chain picture transfer in ...(img-o24VTF6k-1628591532162)]


[ Outside the chain picture transfer in ...(img-KvLHOxFR-1628591532164)]

**[ Data acquisition method ： Get it free of charge here ](https://gitee.com/vip204888/java-p7)**