MySQL之select in 子查詢優(yōu)化的實(shí)現(xiàn)

更新時(shí)間：2020年09月20日 10:59:01 作者：wjw555

這篇文章主要介紹了MySQL之select in 子查詢優(yōu)化的實(shí)現(xiàn)，文中通過示例代碼介紹的非常詳細(xì)，對(duì)大家的學(xué)習(xí)或者工作具有一定的參考學(xué)習(xí)價(jià)值，需要的朋友們下面隨著小編來一起學(xué)習(xí)學(xué)習(xí)吧

下面的演示基于MySQL5.7.27版本

一、關(guān)于MySQL子查詢的優(yōu)化策略介紹:

子查詢優(yōu)化策略

對(duì)于不同類型的子查詢，優(yōu)化器會(huì)選擇不同的策略。

1. 對(duì)于 IN、=ANY 子查詢，優(yōu)化器有如下策略選擇：

semijoin
Materialization
exists

2. 對(duì)于 NOT IN、<>ALL 子查詢，優(yōu)化器有如下策略選擇：

Materialization
exists

3. 對(duì)于 derived 派生表，優(yōu)化器有如下策略選擇：
derived_merge，將派生表合并到外部查詢中（5.7 引入）；
將派生表物化為內(nèi)部臨時(shí)表，再用于外部查詢。
注意：update 和 delete 語句中子查詢不能使用 semijoin、materialization 優(yōu)化策略

二、創(chuàng)建數(shù)據(jù)進(jìn)行模擬演示

為了方便分析問題先建兩張表并插入模擬數(shù)據(jù)：

CREATE TABLE `test02` (
 `id` int(11) NOT NULL,
 `a` int(11) DEFAULT NULL,
 `b` int(11) DEFAULT NULL,
 PRIMARY KEY (`id`),
 KEY `a` (`a`)
) ENGINE=InnoDB;

drop procedure idata;
delimiter ;;
create procedure idata()
begin
 declare i int;
 set i=1;
 while(i<=10000)do
  insert into test02 values(i, i, i);
  set i=i+1;
 end while;
end;;
delimiter ;
call idata();

create table test01 like test02;
insert into test01 (select * from test02 where id<=1000)

三、舉例分析SQL實(shí)例

子查詢示例：

SELECT * FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10)

大部分人可定會(huì)簡單的認(rèn)為這個(gè) SQL 會(huì)這樣執(zhí)行：

SELECT test02.b FROM test02 WHERE id < 10

結(jié)果:1,2,3,4,5,6,7,8,9

SELECT * FROM test01 WHERE test01.a IN (1,2,3,4,5,6,7,8,9);

但實(shí)際上 MySQL 并不是這樣做的。MySQL 會(huì)將相關(guān)的外層表壓到子查詢中，優(yōu)化器認(rèn)為這樣效率更高。也就是說，優(yōu)化器會(huì)將上面的 SQL 改寫成這樣：

select * from test01 where exists(select b from test02 where id < 10 and test01.a=test02.b);

提示：針對(duì)mysql5.5以及之前的版本

查看執(zhí)行計(jì)劃如下，發(fā)現(xiàn)這條SQL對(duì)表test01進(jìn)行了全表掃描1000，效率低下：

root@localhost [dbtest01]>desc select * from test01 where exists(select b from test02 where id < 10 and test01.a=test02.b);
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
| id | select_type    | table | partitions | type | possible_keys | key   | key_len | ref | rows  | filtered | Extra    |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
| 1 | PRIMARY      | test01 | NULL    | ALL  | NULL     | NULL  | NULL  | NULL | 1000  |  100.00 | Using where |
| 2 | DEPENDENT SUBQUERY | test02 | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL |   9 |  10.00 | Using where |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
2 rows in set, 2 warnings (0.00 sec)

但是此時(shí)實(shí)際執(zhí)行下面的SQL，發(fā)現(xiàn)也不慢啊,這不是自相矛盾嘛，別急，咱們繼續(xù)往下分析:

SELECT * FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10)

查看此條SQL的執(zhí)行計(jì)劃如下：

root@localhost [dbtest01]>desc SELECT * FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10);
+----+--------------+-------------+------------+-------+---------------+---------+---------+---------------+------+----------+-------------+
| id | select_type | table    | partitions | type | possible_keys | key   | key_len | ref      | rows | filtered | Extra    |
+----+--------------+-------------+------------+-------+---------------+---------+---------+---------------+------+----------+-------------+
| 1 | SIMPLE    | <subquery2> | NULL    | ALL  | NULL     | NULL  | NULL  | NULL     | NULL |  100.00 | Using where |
| 1 | SIMPLE    | test01   | NULL    | ref  | a       | a    | 5    | <subquery2>.b |  1 |  100.00 | NULL    |
| 2 | MATERIALIZED | test02   | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL     |  9 |  100.00 | Using where |
+----+--------------+-------------+------------+-------+---------------+---------+---------+---------------+------+----------+-------------+
3 rows in set, 1 warning (0.00 sec)

發(fā)現(xiàn)優(yōu)化器使用到了策略MATERIALIZED。于是對(duì)此策略進(jìn)行了資料查詢和學(xué)習(xí)。
https://dev.mysql.com/doc/refman/5.6/en/subquery-optimization.html

原因是從MySQL5.6版本之后包括MySQL5.6版本，優(yōu)化器引入了新的優(yōu)化策略:materialization=[off|on],semijoin=[off|on],(off代表關(guān)閉此策略，on代表開啟此策略)
可以采用show variables like 'optimizer_switch'; 來查看MySQL采用的優(yōu)化器策略。當(dāng)然這些策略都是可以在線進(jìn)行動(dòng)態(tài)修改的
set global optimizer_switch='materialization=on,semijoin=on';代表開啟優(yōu)化策略materialization和semijoin

MySQL5.7.27默認(rèn)的優(yōu)化器策略：

root@localhost [dbtest01]>show variables like 'optimizer_switch';                                                               
+------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Variable_name  | Value                                                                                                                                                                                                      |
+------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| optimizer_switch | index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,engine_condition_pushdown=on,index_condition_pushdown=on,mrr=on,mrr_cost_based=on,block_nested_loop=on,batched_key_access=off,materialization=on,semijoin=on,loosescan=on,firstmatch=on,duplicateweedout=on,subquery_materialization_cost_based=on,use_index_extensions=on,condition_fanout_filter=on,derived_merge=on |
+------------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

所以在MySQL5.6及以上版本時(shí)

執(zhí)行下面的SQL是不會(huì)慢的。因?yàn)镸ySQL的優(yōu)化器策略materialization和semijoin 對(duì)此SQL進(jìn)行了優(yōu)化

SELECT * FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10)

然而咱們把mysql的優(yōu)化器策略materialization和semijoin 關(guān)閉掉測(cè)試，發(fā)現(xiàn)SQL確實(shí)對(duì)test01進(jìn)行了全表的掃描(1000):

set global optimizer_switch='materialization=off,semijoin=off';

執(zhí)行計(jì)劃如下test01表確實(shí)進(jìn)行了全表掃描:

root@localhost [dbtest01]>desc SELECT * FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10);
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
| id | select_type    | table | partitions | type | possible_keys | key   | key_len | ref | rows  | filtered | Extra    |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
| 1 | PRIMARY      | test01 | NULL    | ALL  | NULL     | NULL  | NULL  | NULL | 1000  |  100.00 | Using where |
| 2 | DEPENDENT SUBQUERY | test02 | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL |   9 |  10.00 | Using where |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
2 rows in set, 1 warning (0.00 sec)

下面咱們分析下這個(gè)執(zhí)行計(jì)劃:

?。。?！再次提示:如果是mysql5.5以及之前的版本，或者是mysql5.6以及之后的版本關(guān)閉掉優(yōu)化器策略materialization=off,semijoin=off，得到的SQL執(zhí)行計(jì)劃和下面的是相同的

root@localhost [dbtest01]>desc select * from test01 where exists(select b from test02 where id < 10 and test01.a=test02.b);
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
| id | select_type    | table | partitions | type | possible_keys | key   | key_len | ref | rows | filtered | Extra    |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
| 1 | PRIMARY      | test01 | NULL    | ALL  | NULL     | NULL  | NULL  | NULL | 1000 |  100.00 | Using where |
| 2 | DEPENDENT SUBQUERY | test02 | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL |  9 |  10.00 | Using where |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+------+----------+-------------+
2 rows in set, 2 warnings (0.00 sec)

不相關(guān)子查詢變成了關(guān)聯(lián)子查詢（select_type:DEPENDENT SUBQUERY），子查詢需要根據(jù) b 來關(guān)聯(lián)外表 test01，因?yàn)樾枰獗淼?test01 字段，所以子查詢是沒法先執(zhí)行的。執(zhí)行流程為：

掃描 test01，從 test01 取出一行數(shù)據(jù) R；
從數(shù)據(jù)行 R 中，取出字段 a 執(zhí)行子查詢，如果得到結(jié)果為 TRUE，則把這行數(shù)據(jù) R 放到結(jié)果集；
重復(fù) 1、2 直到結(jié)束。

總的掃描行數(shù)為 1000+1000*9=10000（這是理論值，但是實(shí)際值比10000還少，怎么來的一直沒想明白，看規(guī)律是子查詢結(jié)果集每多一行，總掃描行數(shù)就會(huì)少幾行）。

Semi-join優(yōu)化器:

這樣會(huì)有個(gè)問題，如果外層表是一個(gè)非常大的表，對(duì)于外層查詢的每一行，子查詢都得執(zhí)行一次，這個(gè)查詢的性能會(huì)非常差。我們很容易想到將其改寫成 join 來提升效率：

select test01.* from test01 join test02 on test01.a=test02.b and test02.id<10;

# 查看此SQL的執(zhí)行計(jì)劃:

desc select test01.* from test01 join test02 on test01.a=test02.b and test02.id<10;

root@localhost [dbtest01]>EXPLAIN extended select test01.* from test01 join test02 on test01.a=test02.b and test02.id<10;
+----+-------------+--------+------------+-------+---------------+---------+---------+-------------------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key   | key_len | ref        | rows | filtered | Extra    |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------------------+------+----------+-------------+
| 1 | SIMPLE   | test02 | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL       |  9 |  100.00 | Using where |
| 1 | SIMPLE   | test01 | NULL    | ref  | a       | a    | 5    | dbtest01.test02.b |  1 |  100.00 | NULL    |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------------------+------+----------+-------------+
2 rows in set, 2 warnings (0.00 sec)

這樣優(yōu)化可以讓 t2 表做驅(qū)動(dòng)表，t1 表關(guān)聯(lián)字段有索引，查找效率非常高。

但這里會(huì)有個(gè)問題，join 是有可能得到重復(fù)結(jié)果的，而 in(select ...) 子查詢語義則不會(huì)得到重復(fù)值。
而 semijoin 正是解決重復(fù)值問題的一種特殊聯(lián)接。
在子查詢中，優(yōu)化器可以識(shí)別出 in 子句中每組只需要返回一個(gè)值，在這種情況下，可以使用 semijoin 來優(yōu)化子查詢，提升查詢效率。
這是 MySQL 5.6 加入的新特性，MySQL 5.6 以前優(yōu)化器只有 exists 一種策略來“優(yōu)化”子查詢。

經(jīng)過 semijoin 優(yōu)化后的 SQL 和執(zhí)行計(jì)劃分為：

root@localhost [dbtest01]>desc SELECT * FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10);
+----+--------------+-------------+------------+-------+---------------+---------+---------+---------------+------+----------+-------------+
| id | select_type | table    | partitions | type | possible_keys | key   | key_len | ref      | rows | filtered | Extra    |
+----+--------------+-------------+------------+-------+---------------+---------+---------+---------------+------+----------+-------------+
| 1 | SIMPLE    | <subquery2> | NULL    | ALL  | NULL     | NULL  | NULL  | NULL     | NULL |  100.00 | Using where |
| 1 | SIMPLE    | test01   | NULL    | ref  | a       | a    | 5    | <subquery2>.b |  1 |  100.00 | NULL    |
| 2 | MATERIALIZED | test02   | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL     |  9 |  100.00 | Using where |
+----+--------------+-------------+------------+-------+---------------+---------+---------+---------------+------+----------+-------------+
3 rows in set, 1 warning (0.00 sec)

select 
  `test01`.`id`,`test01`.`a`,`test01`.`b` 
from `test01` semi join `test02` 
where
  ((`test01`.`a` = `<subquery2>`.`b`) 
  and (`test02`.`id` < 10));

##注意這是優(yōu)化器改寫的SQL，客戶端上是不能用 semi join 語法的

semijoin 優(yōu)化實(shí)現(xiàn)比較復(fù)雜，其中又分 FirstMatch、Materialize 等策略,上面的執(zhí)行計(jì)劃中 select_type=MATERIALIZED 就是代表使用了 Materialize 策略來實(shí)現(xiàn)的 semijoin
這里 semijoin 優(yōu)化后的執(zhí)行流程為：

先執(zhí)行子查詢，把結(jié)果保存到一個(gè)臨時(shí)表中，這個(gè)臨時(shí)表有個(gè)主鍵用來去重；
從臨時(shí)表中取出一行數(shù)據(jù) R；
從數(shù)據(jù)行 R 中，取出字段 b 到被驅(qū)動(dòng)表 t1 中去查找，滿足條件則放到結(jié)果集；
重復(fù)執(zhí)行 2、3，直到結(jié)束。
這樣一來，子查詢結(jié)果有 9 行，即臨時(shí)表也有 9 行（這里沒有重復(fù)值），總的掃描行數(shù)為 9+9+9*1=27 行，比原來的 10000 行少了很多。

MySQL 5.6 版本中加入的另一種優(yōu)化特性 materialization，就是把子查詢結(jié)果物化成臨時(shí)表，然后代入到外查詢中進(jìn)行查找，來加快查詢的執(zhí)行速度。內(nèi)存臨時(shí)表包含主鍵（hash 索引），消除重復(fù)行，使表更小。
如果子查詢結(jié)果太大，超過 tmp_table_size 大小，會(huì)退化成磁盤臨時(shí)表。這樣子查詢只需要執(zhí)行一次，而不是對(duì)于外層查詢的每一行都得執(zhí)行一遍。
不過要注意的是，這樣外查詢依舊無法通過索引快速查找到符合條件的數(shù)據(jù)，只能通過全表掃描或者全索引掃描，

semijoin 和 materialization 的開啟是通過 optimizer_switch 參數(shù)中的 semijoin={on|off}、materialization={on|off} 標(biāo)志來控制的。
上文中不同的執(zhí)行計(jì)劃就是對(duì) semijoin 和 materialization 進(jìn)行開/關(guān)產(chǎn)生的
總的來說對(duì)于子查詢，先檢查是否滿足各種優(yōu)化策略的條件（比如子查詢中有 union 則無法使用 semijoin 優(yōu)化）
然后優(yōu)化器會(huì)按成本進(jìn)行選擇，實(shí)在沒得選就會(huì)用 exists 策略來“優(yōu)化”子查詢，exists 策略是沒有參數(shù)來開啟或者關(guān)閉的。

下面舉一個(gè)delete相關(guān)的子查詢例子：

把上面的2張測(cè)試表分別填充350萬數(shù)據(jù)和50萬數(shù)據(jù)來測(cè)試delete語句

root@localhost [dbtest01]>select count(*) from test02;
+----------+
| count(*) |
+----------+
| 3532986 |
+----------+
1 row in set (0.64 sec)
root@localhost [dbtest01]>create table test01 like test02;
Query OK, 0 rows affected (0.01 sec)

root@localhost [dbtest01]>insert into test01 (select * from test02 where id<=500000)

root@localhost [dbtest01]>select count(*) from test01;
+----------+
| count(*) |
+----------+
|  500000 |

執(zhí)行delete刪除語句執(zhí)行了4s

root@localhost [dbtest01]>delete FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10);
Query OK, 9 rows affected (4.86 sec)

查看執(zhí)行計(jì)劃，對(duì)test01表進(jìn)行了幾乎全表掃描:

root@localhost [dbtest01]>desc delete FROM test01 WHERE test01.a IN (SELECT test02.b FROM test02 WHERE id < 10);
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
| id | select_type    | table | partitions | type | possible_keys | key   | key_len | ref | rows  | filtered | Extra    |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
| 1 | DELETE       | test01 | NULL    | ALL  | NULL     | NULL  | NULL  | NULL | 499343 |  100.00 | Using where |
| 2 | DEPENDENT SUBQUERY | test02 | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL |   9 |  10.00 | Using where |
+----+--------------------+--------+------------+-------+---------------+---------+---------+------+--------+----------+-------------+
2 rows in set (0.00 sec)

于是修改上面的delete SQL語句偽join語句

root@localhost [dbtest01]>desc delete test01.* from test01 join test02 on test01.a=test02.b and test02.id<10;
+----+-------------+--------+------------+-------+---------------+---------+---------+-------------------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key   | key_len | ref        | rows | filtered | Extra    |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------------------+------+----------+-------------+
| 1 | SIMPLE   | test02 | NULL    | range | PRIMARY    | PRIMARY | 4    | NULL       |  9 |  100.00 | Using where |
| 1 | DELETE   | test01 | NULL    | ref  | a       | a    | 5    | dbtest01.test02.b |  1 |  100.00 | NULL    |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------------------+------+----------+-------------+
2 rows in set (0.01 sec)

執(zhí)行非常的快
root@localhost [dbtest01]>delete test01.* from test01 join test02 on test01.a=test02.b and test02.id<10;
Query OK, 9 rows affected (0.01 sec)

root@localhost [dbtest01]>select test01.* from test01 join test02 on test01.a=test02.b and test02.id<10;
Empty set (0.00 sec)

下面的這個(gè)表執(zhí)行要全表掃描，非常慢，基本對(duì)表test01進(jìn)行了全表掃描:

root@lcalhost [dbtest01]>desc delete FROM test01 WHERE id IN (SELECT id FROM test02 WHERE id='350000');
+----+--------------------+--------+------------+-------+---------------+---------+---------+-------+--------+----------+-------------+
| id | select_type    | table | partitions | type | possible_keys | key   | key_len | ref  | rows  | filtered | Extra    |
+----+--------------------+--------+------------+-------+---------------+---------+---------+-------+--------+----------+-------------+
| 1 | DELETE       | test01 | NULL    | ALL  | NULL     | NULL  | NULL  | NULL | 499343 |  100.00 | Using where |
| 2 | DEPENDENT SUBQUERY | test02 | NULL    | const | PRIMARY    | PRIMARY | 4    | const |   1 |  100.00 | Using index |
+----+--------------------+--------+------------+-------+---------------+---------+---------+-------+--------+----------+-------------+
2 rows in set (0.00 sec)

然而采用join的話，效率非常的高:

root@localhost [dbtest01]>desc delete test01.* FROM test01 inner join test02 WHERE test01.id=test02.id and test02.id=350000 ;
+----+-------------+--------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key   | key_len | ref  | rows | filtered | Extra    |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
| 1 | DELETE   | test01 | NULL    | const | PRIMARY    | PRIMARY | 4    | const |  1 |  100.00 | NULL    |
| 1 | SIMPLE   | test02 | NULL    | const | PRIMARY    | PRIMARY | 4    | const |  1 |  100.00 | Using index |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------+------+----------+-------------+
2 rows in set (0.01 sec)

 
root@localhost [dbtest01]> desc delete test01.* from test01 join test02 on test01.a=test02.b and test02.id=350000;
+----+-------------+--------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key   | key_len | ref  | rows | filtered | Extra |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
| 1 | SIMPLE   | test02 | NULL    | const | PRIMARY    | PRIMARY | 4    | const |  1 |  100.00 | NULL |
| 1 | DELETE   | test01 | NULL    | ref  | a       | a    | 5    | const |  1 |  100.00 | NULL |
+----+-------------+--------+------------+-------+---------------+---------+---------+-------+------+----------+-------+
2 rows in set (0.00 sec)

參考文檔:

https://www.cnblogs.com/zhengyun_ustc/p/slowquery1.html
https://www.jianshu.com/p/3989222f7084
https://dev.mysql.com/doc/refman/5.6/en/subquery-optimization.html

到此這篇關(guān)于MySQL之select in 子查詢優(yōu)化的實(shí)現(xiàn)的文章就介紹到這了,更多相關(guān)MySQL select in 子查詢優(yōu)化內(nèi)容請(qǐng)搜索腳本之家以前的文章或繼續(xù)瀏覽下面的相關(guān)文章希望大家以后多多支持腳本之家！

您可能感興趣的文章: