Mysql Join大表小表到底谁驱动谁
Posted valleyup
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了Mysql Join大表小表到底谁驱动谁相关的知识，希望对你有一定的参考价值。
1、准备
mysql> create table dept( id int unsigned auto_increment not null primary key, name varchar(20) default ‘‘ not null, key(name) )engine=innodb default charset=utf8mb4;
CREATE TABLE `userinfo` ( `id` int NOT NULL AUTO_INCREMENT, `name` varchar(255) COLLATE utf8mb4_general_ci NOT NULL, `passwd` varchar(255) COLLATE utf8mb4_general_ci DEFAULT NULL, `phone` varchar(255) COLLATE utf8mb4_general_ci DEFAULT NULL, `dept` int NOT NULL DEFAULT ‘0‘, PRIMARY KEY (`id`), KEY `name` (`name`), KEY `dept` (`dept`) ) ENGINE=InnoDB AUTO_INCREMENT=1001 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci
用 excel 伪造数据，并导入
mysql> load data infile ‘/var/lib/mysql-files/new5.txt‘ into table dept fields terminated by ‘ ‘ lines terminated by ‘ ‘; Query OK, 1000 rows affected (0.03 sec) Records: 1000 Deleted: 0 Skipped: 0 Warnings: 0 mysql> load data infile ‘/var/lib/mysql-files/new.txt‘ into table userinfo fields terminated by ‘ ‘ lines terminated by ‘ ‘; Query OK, 120000 rows affected (0.02 sec) Records: 120000 Deleted: 0 Skipped: 0 Warnings: 0
dept 1000 条， userinfo 120000 条
一、查找 userinfo 名字是 a25 的 id，name，及 dept name
由于有 userinfo = ‘a25‘ 的限制条件，不管谁 join 谁，或者 left join，结果都是一样的
mysql> select count(1) from userinfo left join dept using(dept_id) where userinfo.name = ‘a25‘; +----------+ | count(1) | +----------+ | 187 | +----------+ 1 row in set mysql> select count(1) from dept left join userinfo using(dept_id) where userinfo.name = ‘a25‘; +----------+ | count(1) | +----------+ | 187 | +----------+ 1 row in set mysql> select count(1) from dept join userinfo using(dept_id) where userinfo.name = ‘a25‘; +----------+ | count(1) | +----------+ | 187 | +----------+ 1 row in set mysql> select count(1) from userinfo join dept using(dept_id) where userinfo.name = ‘a25‘; +----------+ | count(1) | +----------+ | 187 | +----------+ 1 row in set
那就比较四种方式的效率：
开启慢查询：
set long_query_time = 0.0001;
set global slow_query_log = ON;
mysql> show variables like "slow%"; +---------------------+--------------------------------------+ | Variable_name | Value | +---------------------+--------------------------------------+ | slow_launch_time | 2 | | slow_query_log | ON | | slow_query_log_file | /var/lib/mysql/muqing-web-2-slow.log | +---------------------+--------------------------------------+
tail -f /var/lib/mysql/muqing-web-2-slow.log // 实时查看耗时
1、userinfo left join dept
mysql> explain select userinfo.user_id, userinfo.name, dept.name from userinfo left join dept using(dept_id) where userinfo.name = ‘a25‘; +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | 1 | SIMPLE | userinfo | NULL | ref | name,name_2 | name | 1022 | const | 187 | 100 | NULL | | 1 | SIMPLE | dept | NULL | eq_ref | PRIMARY | PRIMARY | 4 | test.userinfo.dept_id | 1 | 100 | Using where | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+------------+ | Variable_name | Value | +-----------------+------------+ | Last_query_cost | 130.899000 | +-----------------+------------+ 1 row in set mysql> explain select straight_join userinfo.user_id, userinfo.name, dept.name from userinfo left join dept using(dept_id) where userinfo.name = ‘a25‘; +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | 1 | SIMPLE | userinfo | NULL | ref | name,name_2 | name | 1022 | const | 187 | 100 | NULL | | 1 | SIMPLE | dept | NULL | eq_ref | PRIMARY | PRIMARY | 4 | test.userinfo.dept_id | 1 | 100 | Using where | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+------------+ | Variable_name | Value | +-----------------+------------+ | Last_query_cost | 130.899000 | +-----------------+------------+ 1 row in set
可以看到优化器选择的是顺序是先 userinfo 表，warnings 信息：
mysql> explain select straight_join userinfo.user_id, userinfo.name, dept.name from userinfo left join dept using(dept_id) where userinfo.name = ‘a25‘; +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | 1 | SIMPLE | userinfo | NULL | ref | name,name_2 | name | 1022 | const | 187 | 100 | NULL | | 1 | SIMPLE | dept | NULL | eq_ref | PRIMARY | PRIMARY | 4 | test.userinfo.dept_id | 1 | 100 | Using where | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ 2 rows in set mysql> show warnings; +-------+------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | Level | Code | Message | +-------+------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | Note | 1003 | /* select#1 */ select straight_join `test`.`userinfo`.`user_id` AS `user_id`,`test`.`userinfo`.`name` AS `name`,`test`.`dept`.`name` AS `name` from `test`.`userinfo` left join `test`.`dept` on((`test`.`userinfo`.`dept_id` = `test`.`dept`.`dept_id`)) where (`test`.`userinfo`.`name` = ‘a25‘) | +-------+------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 1 row in set
2、userinfo join dept
mysql> explain select userinfo.user_id, userinfo.name, dept.name from userinfo join dept using(dept_id) where userinfo.name = ‘a25‘; +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | 1 | SIMPLE | userinfo | NULL | ref | name,name_2 | name | 1022 | const | 187 | 100 | NULL | | 1 | SIMPLE | dept | NULL | eq_ref | PRIMARY | PRIMARY | 4 | test.userinfo.dept_id | 1 | 100 | Using where | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+------------+ | Variable_name | Value | +-----------------+------------+ | Last_query_cost | 130.899000 | +-----------------+------------+ 1 row in set mysql> explain select straight_join userinfo.user_id, userinfo.name, dept.name from userinfo join dept using(dept_id) where userinfo.name = ‘a25‘; +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ | 1 | SIMPLE | userinfo | NULL | ref | name,name_2 | name | 1022 | const | 187 | 100 | NULL | | 1 | SIMPLE | dept | NULL | eq_ref | PRIMARY | PRIMARY | 4 | test.userinfo.dept_id | 1 | 100 | Using where | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+------+----------+-------------+ 2 rows in set mysql> show warnings; +-------+------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | Level | Code | Message | +-------+------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | Note | 1003 | /* select#1 */ select straight_join `test`.`userinfo`.`user_id` AS `user_id`,`test`.`userinfo`.`name` AS `name`,`test`.`dept`.`name` AS `name` from `test`.`userinfo` join `test`.`dept` where ((`test`.`userinfo`.`name` = ‘a25‘) and (`test`.`userinfo`.`dept_id` = `test`.`dept`.`dept_id`)) | +-------+------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 1 row in set
# Time: 2020-04-10T17:00:38.452393Z # User@Host: root[root] @ localhost [] Id: 11 # Query_time: 0.000741 Lock_time: 0.000137 Rows_sent: 187 Rows_examined: 374 SET timestamp=1586538038; select sql_no_cache userinfo.user_id, userinfo.name, dept.name from userinfo left join dept using(dept_id) where userinfo.name = ‘a25‘; # Time: 2020-04-10T17:01:49.946904Z # User@Host: root[root] @ localhost [] Id: 11 # Query_time: 0.000766 Lock_time: 0.000149 Rows_sent: 187 Rows_examined: 374 SET timestamp=1586538109; select straight_join userinfo.user_id, userinfo.name, dept.name from userinfo join dept using(dept_id) where userinfo.name = ‘a25‘;
可以看出两个时间差不多
3、dept left join userinfo
mysql> explain select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept left join userinfo using(dept_id) where userinfo.name = ‘a25‘; +----+-------------+----------+------------+-------+---------------+------+---------+-------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+-------+---------------+------+---------+-------+------+----------+-------------+ | 1 | SIMPLE | dept | NULL | index | PRIMARY | name | 82 | NULL | 1000 | 100 | Using index | | 1 | SIMPLE | userinfo | NULL | ref | name,name_2 | name | 1022 | const | 187 | 10 | Using where | +----+-------------+----------+------------+-------+---------------+------+---------+-------+------+----------+-------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+---------------+ | Variable_name | Value | +-----------------+---------------+ | Last_query_cost | 106671.417848 | +-----------------+---------------+ 1 row in set
# Time: 2020-04-10T17:32:02.890446Z # User@Host: root[root] @ [10.0.0.105] Id: 13 # Query_time: 0.134147 Lock_time: 0.000128 Rows_sent: 187 Rows_examined: 188000 SET timestamp=1586539922; select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept left join userinfo using(dept_id) where userinfo.name = ‘a25‘;
明显比上面的要慢
4、dept join userinfo
mysql> explain select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept join userinfo using(dept_id) where userinfo.name = ‘a25‘; +----+-------------+----------+------------+-------+---------------+------+---------+-------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+-------+---------------+------+---------+-------+------+----------+-------------+ | 1 | SIMPLE | dept | NULL | index | PRIMARY | name | 82 | NULL | 1000 | 100 | Using index | | 1 | SIMPLE | userinfo | NULL | ref | name,name_2 | name | 1022 | const | 187 | 10 | Using where | +----+-------------+----------+------------+-------+---------------+------+---------+-------+------+----------+-------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+---------------+ | Variable_name | Value | +-----------------+---------------+ | Last_query_cost | 106671.417848 | +-----------------+---------------+ 1 row in set
# Time: 2020-04-10T17:33:35.458204Z # User@Host: root[root] @ [10.0.0.105] Id: 13 # Query_time: 0.133475 Lock_time: 0.000164 Rows_sent: 187 Rows_examined: 188000 SET timestamp=1586540015; select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept join userinfo using(dept_id) where userinfo.name = ‘a25‘;
速度和 3 一样，也远大于前2
所以是大表驱动小表。而且 show status like ‘last_query_cost‘ 还挺准的
二、查找 dept 名字是 a25 的 id，name，及 dept name
四种查询等效
mysql> select count(1) from userinfo left join dept using(dept_id) where dept.name = ‘a25‘; +----------+ | count(1) | +----------+ | 360 | +----------+ 1 row in set mysql> select count(1) from userinfo join dept using(dept_id) where dept.name = ‘a25‘; +----------+ | count(1) | +----------+ | 360 | +----------+ 1 row in set mysql> select count(1) from dept join userinfo using(dept_id) where dept.name = ‘a25‘; +----------+ | count(1) | +----------+ | 360 | +----------+ 1 row in set mysql> select count(1) from dept left join userinfo using(dept_id) where dept.name = ‘a25‘; +----------+ | count(1) | +----------+ | 360 | +----------+ 1 row in set
1、userinfo left join dept
mysql> explain select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from userinfo left join dept using(dept_id) where dept.name = ‘a25‘; +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+--------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+--------+----------+-------------+ | 1 | SIMPLE | userinfo | NULL | ALL | NULL | NULL | NULL | NULL | 120234 | 100 | NULL | | 1 | SIMPLE | dept | NULL | eq_ref | PRIMARY,name | PRIMARY | 4 | test.userinfo.dept_id | 1 | 5 | Using where | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+--------+----------+-------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+--------------+ | Variable_name | Value | +-----------------+--------------+ | Last_query_cost | 54209.549000 | +-----------------+--------------+ 1 row in set
# Time: 2020-04-10T17:40:53.462276Z # User@Host: root[root] @ [10.0.0.105] Id: 13 # Query_time: 0.110679 Lock_time: 0.000145 Rows_sent: 360 Rows_examined: 240000 SET timestamp=1586540453; select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from userinfo left join dept using(dept_id) where dept.name = ‘a25‘;
2、userinfo join dept
mysql> explain select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from userinfo join dept using(dept_id) where dept.name = ‘a25‘; +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+--------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+--------+----------+-------------+ | 1 | SIMPLE | userinfo | NULL | ALL | NULL | NULL | NULL | NULL | 120234 | 100 | NULL | | 1 | SIMPLE | dept | NULL | eq_ref | PRIMARY,name | PRIMARY | 4 | test.userinfo.dept_id | 1 | 5 | Using where | +----+-------------+----------+------------+--------+---------------+---------+---------+-----------------------+--------+----------+-------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+--------------+ | Variable_name | Value | +-----------------+--------------+ | Last_query_cost | 54209.549000 | +-----------------+--------------+
# Time: 2020-04-10T17:44:21.701356Z # User@Host: root[root] @ [10.0.0.105] Id: 13 # Query_time: 0.105262 Lock_time: 0.000137 Rows_sent: 360 Rows_examined: 240000 SET timestamp=1586540661; select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from userinfo join dept using(dept_id) where dept.name = ‘a25‘;
速度比上面的好一点点
3、dept left join userinfo
mysql> explain select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept left join userinfo using(dept_id) where dept.name = ‘a25‘; +----+-------------+----------+------------+------+---------------+------+---------+-------+--------+----------+----------------------------------------------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+------+---------------+------+---------+-------+--------+----------+----------------------------------------------------+ | 1 | SIMPLE | dept | NULL | ref | name | name | 82 | const | 1 | 100 | Using index | | 1 | SIMPLE | userinfo | NULL | ALL | NULL | NULL | NULL | NULL | 120234 | 100 | Using where; Using join buffer (Block Nested Loop) | +----+-------------+----------+------------+------+---------------+------+---------+-------+--------+----------+----------------------------------------------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+--------------+ | Variable_name | Value | +-----------------+--------------+ | Last_query_cost | 12128.032803 | +-----------------+--------------+
# Time: 2020-04-10T17:47:06.147081Z # User@Host: root[root] @ [10.0.0.105] Id: 13 # Query_time: 0.031848 Lock_time: 0.000137 Rows_sent: 360 Rows_examined: 120001 SET timestamp=1586540826; select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept left join userinfo using(dept_id) where dept.name = ‘a25‘;
速度竟然比上面的块，不是大表驱动小表吗，继续看
4、dept join userinfo
mysql> explain select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept join userinfo using(dept_id) where dept.name = ‘a25‘; +----+-------------+----------+------------+------+---------------+------+---------+-------+--------+----------+----------------------------------------------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+----------+------------+------+---------------+------+---------+-------+--------+----------+----------------------------------------------------+ | 1 | SIMPLE | dept | NULL | ref | PRIMARY,name | name | 82 | const | 1 | 100 | Using index | | 1 | SIMPLE | userinfo | NULL | ALL | NULL | NULL | NULL | NULL | 120234 | 10 | Using where; Using join buffer (Block Nested Loop) | +----+-------------+----------+------------+------+---------------+------+---------+-------+--------+----------+----------------------------------------------------+ 2 rows in set mysql> show status like ‘last_query_cost‘; +-----------------+--------------+ | Variable_name | Value | +-----------------+--------------+ | Last_query_cost | 12128.032803 | +-----------------+--------------+
# Time: 2020-04-10T17:48:44.738339Z # User@Host: root[root] @ [10.0.0.105] Id: 13 # Query_time: 0.032005 Lock_time: 0.000143 Rows_sent: 360 Rows_examined: 120001 SET timestamp=1586540924; select sql_no_cache straight_join userinfo.user_id, userinfo.name, dept.name from dept join userinfo using(dept_id) where dept.name = ‘a25‘;
速度也比 1，2快
left join 和 join 相比 按理说会慢一点，因为：join 的伪代码为
outer_iter = iterator over tb1 where col1 in(5, 6) outer_row = outer_iter.next while outer_row inner_iter = iterator over tb2 where col3 = outer_row.col3 inner_row = inner_iter.next while inner_row out_put [ outer_row.col1, inner_row.col2 ] inner_row = inner_iter.next end outer_row = outer_iter.next end
而 left join 会多一次判断，多一次返回 null 结果集。
猜想是 where 条件 获取的过滤多，就用谁来做外表，所以讲不定 一定是大表驱动小表
以上是关于Mysql Join大表小表到底谁驱动谁的主要内容，如果未能解决你的问题，请参考以下文章