当前位置：网站首页>Xiaobai learns MySQL - derived table

Xiaobai learns MySQL - derived table

2022-07-26 22:17:00 【bisal(Chen Liu)】

A friend recently asked this question ,MySQL one SQL Implementation plan , As shown below , Among them is PRIMARY、<derived2>、DERIVED These content , How did he and SQL Corresponding upper ？

MySQL There are indeed some and Oracle Different technical terms , But the principle and mechanism behind it are interlinked .

To illustrate , Simulate the creation of test tables ,

create table t01(
  id int,
  code varchar(10),
  start_date datetime,
  title varchar(10),
  content varchar(30),
  class int,
  end_date datetime
);

Insert some data ,

[email protected]:  [test]> select * from t01;
+------+--------+---------------------+---------+------------+-------+---------------------+
| id   | code   | start_date          | title   | content    | class | end_date            |
+------+--------+---------------------+---------+------------+-------+---------------------+
|    1 | code1  | 2022-01-01 00:00:00 | title1  | content_1  |     1 | 2022-01-03 00:00:00 |
|    2 | code2  | 2022-01-02 00:00:00 | title2  | content_2  |     3 | 2022-01-03 00:00:00 |
|    3 | code3  | 2022-01-03 00:00:00 | title3  | content_3  |     2 | 2022-01-03 00:00:00 |
|    4 | code4  | 2022-01-04 00:00:00 | title4  | content_4  |     1 | 2022-01-06 00:00:00 |
|    5 | code5  | 2022-01-05 00:00:00 | title5  | content_5  |     1 | 2022-01-07 00:00:00 |
|    6 | code6  | 2022-01-06 00:00:00 | title6  | content_6  |     2 | 2022-01-10 00:00:00 |
|    7 | code7  | 2022-01-07 00:00:00 | title7  | content_7  |     1 | 2022-01-11 00:00:00 |
|    8 | code8  | 2022-01-08 00:00:00 | title8  | content_8  |     1 | 2022-01-12 00:00:00 |
|    9 | code9  | 2022-01-09 00:00:00 | title9  | content_9  |     3 | 2022-01-10 00:00:00 |
|   10 | code10 | 2022-01-10 00:00:00 | title10 | content_10 |     1 | 2022-01-13 00:00:00 |
+------+--------+---------------------+---------+------------+-------+---------------------+
10 rows in set (0.00 sec)

The implementation is this SQL,

[email protected]:  [test]>
    -> select * from (
    -> (select id, code, start_date, title, content, concat(' test 1:', start_date), class, end_date
    -> from t01 t
    -> where id=1 and code='code1' and title='title1' and start_date>='2022-01-01' and start_date<='2022-01-05'
    -> order by end_date desc limit 1)
    -> union all
    -> (select id, code, start_date, title, content, concat(' test 2:', start_date), class, end_date
    -> from t01 t
    -> where id=2 and code='code2' and title='title2' and start_date>='2022-01-01' and start_date<='2022-01-05'
    -> order by end_date desc limit 1)
    -> union all
    -> (select id, code, start_date, title, content, concat(' test 3:', start_date), class, end_date
    -> from t01 t
    -> where id=3 and code='code3' and title='title3' and start_date>='2022-01-01' and start_date<='2022-01-05'
    -> order by end_date desc limit 1)
    -> union all
    -> (select id, code, start_date, title, content, concat(' test 4:', title), class, end_date
    -> from t01 t
    -> where id=4 and code='code4' and title='title4' and start_date>='2022-01-01' and start_date<='2022-01-05' and end_date<='2022-01-05')
    -> union all
    -> (select id, code, start_date, title, content, concat(' test 5:', content), class, end_date
    -> from t01 t
    -> where id=5 and code='code5' and title='title5' and start_date>='2022-01-01' and start_date<='2022-01-05'and end_date<='2022-01-05')
    -> ) tt order by class, end_date desc;

His execution plan , It is basically the same as what was mentioned at the beginning of the article ,

From the official documents , We can see , What is? Derived Tables？ simply , It will be FROM The search result set appearing in the clause is treated as a table , for example FROM Medium SELECT Subquery is a derived table, And each FROM All tables in the clause need a table alias , Any from derived table The column of must have a unique name , Other requirements and examples , You can refer to the link ,

https://dev.mysql.com/doc/refman/5.7/en/derived-tables.html

A derived table is an expression that generates a table within the scope of a query FROM clause. For example, a subquery in a SELECT statement FROM clause is a derived table:
SELECT … FROM (subquery) [AS] tbl_name
The [AS] tbl_name clause is mandatory because every table in a FROM clause must have a name. Any columns in the derived table must have unique names

Execute the first line in the plan <derived2> This piece of " surface " It's a full table scan ,

The reason is the above SQL, Actually, it can be understood as , Yes derived table Search for , In fact, there are no search conditions ,

select * from ( ... ) tt order by class, end_date desc;

In fact, carefully observe the above SQL,derived table in union all The first three connected SQL The search conditions are basically the same , and union all The last two connected SQL The search conditions are basically the same , It's just SELECT in concat The content is different , So I can rewrite .

The previous three SQL For example ,concat adopt case when Judge different id and title Under the condition of , What should be output ,where Bring all the previous fields in the condition , After the transformation, this is an independent SQL, There are no subqueries ,

[email protected]:  [test]> explain
    -> select id, code, start_date, content,
    ->        (case when id=1 and code='code1' and title='title1' then concat(' test 1:', start_date)
    ->              when id=2 and code='code2' and title='title2' then concat(' test 2:', start_date)
    ->              when id=3 and code='code3' and title='title3' then concat(' test 3:', start_date) end) c,
    ->        class, end_date
    -> from t01
    -> where id in (1, 2, 3) and title in ('title1', 'title2', 'title3')
    -> and start_date>='2022-01-01' and start_date<='2022-01-05'
    -> order by class, end_date desc;
+----+-------------+-------+------------+-------+---------------+------------+---------+------+------+----------+----------------------------------------------------+
| id | select_type | table | partitions | type  | possible_keys | key        | key_len | ref  | rows | filtered | Extra                                              |
+----+-------------+-------+------------+-------+---------------+------------+---------+------+------+----------+----------------------------------------------------+
|  1 | SIMPLE      | t01   | NULL       | range | idx_t01_01    | idx_t01_01 | 5       | NULL |    3 |    10.00 | Using index condition; Using where; Using filesort |
+----+-------------+-------+------------+-------+---------------+------------+---------+------+------+----------+----------------------------------------------------+
1 row in set, 1 warning (0.00 sec)

One of the optimization rules , Just do less . If rewritten as this , The biggest advantage , The same table only needs to be read once , Before derived table Each of them union all All sub queries of need to read the table once t01.

Xiaobai studies MySQL

《 Xiaobai studies MySQL - Generated Columns function 》

《 Xiaobai studies MySQL - Incremental Statistics SQL The needs of - Scheme of windowed function 》

《 Xiaobai studies MySQL - The statistical " Be opportunistic "》

《 Xiaobai studies MySQL - Incremental Statistics SQL The needs of 》

《 Xiaobai studies MySQL - You've encountered this kind of scenario where you can't log in ？》

《 Xiaobai studies MySQL - There are some differences between users created by different versions 》

《 Xiaobai studies MySQL - A tool for randomly inserting test data 》

《 Xiaobai studies MySQL - varchar Why are type fields often defined as 255？》

《 Xiaobai studies MySQL - A case of flexible index creation 》

《 Xiaobai studies MySQL - “ Be opportunistic ” The number of records in a statistical table 》

《 Xiaobai studies MySQL - Once slow SQL The positioning of 》

《 Xiaobai studies MySQL - TIMESTAMP The type field is not empty and the default value property is affected 》

《 Xiaobai studies MySQL - Talk about the importance of data backup 》

《 Xiaobai studies MySQL - InnoDB Support optimize table？》

《 Xiaobai studies MySQL - table_open_cache The role of 》

《 Xiaobai studies MySQL - Table space defragmentation method 》

《 Xiaobai studies MySQL - Case sensitive problem solving 》

《 Xiaobai studies MySQL - only_full_group_by Validation rules for 》

《 Xiaobai studies MySQL - max_allowed_packet》

《 Xiaobai studies MySQL - mysqldump Parameter differences to ensure data consistency 》

《 Xiaobai studies MySQL - The query will lock the table ？》

《 Xiaobai studies MySQL - The problem of index key length limitation 》

《 Xiaobai studies MySQL - MySQL Will be affected by “ High water level ” Influence ？》

《 Xiaobai studies MySQL - Database software and initialization installation 》

《 Xiaobai studies MySQL - Chat 》

If you think this article is helpful , Please also click at the end of the article " give the thumbs-up " and " Looking at ", Or forward it directly pyq,