本文同时发表在https://github.com/zhangyachen/zhangyachen.github.io/issues/51
当前mysql执行的策略很简单：mysql对任何关联都执行嵌套循环操作，即mysql先在一个表中循环取出单条数据，然后再嵌套循环到下一个表中寻打匹配的行，依次下去，直到描述到所表表中匹配的行为止。然后根据各个表匹配的行，返回查询中需要的各个列。mysql会尝试在最后一个关联表中打到所有匹配的行，如果最后一个关联表无法找到更多的行以后，mysql返回到上一层次关联表，看是否能够找到更多的匹配记录，依此类推迭代执行。
按照这样的方式查找第一条表记录，再嵌套查询下一个关联表，然后回溯到上一个表，在mysql中是通过嵌套循环的方式来实现的--正如其名‘嵌套循环关联’。请看下面的例子中的简单查询：
select tbl1.col1,tbl2.col2
from tbl1 inner join tbl2 using(col3)
where tbl1.col1 in(5,6);
假设mysql按照查询中的表顺序进行关联操作，我们则可以用下面的伪代码表示mysql将如何完成这个查询：
outer_iter = iterator_over tbl1 where col1 in(3,4)
outer_row = outer_iter.next
while outer_row
inner_iter = iterator over tbl2 where col3=outer_row.col3
inner_row = inner_iter.next
while inner_row
output[outer_row.col1,inner_row.col2]
inner_row = inner_iter.next
end
out_row = outer_iter.next
end
上面的执行计划对于单表查询和多表关联查询都适用，如果是一个单表查询，那么只需要完成上面的外层的基本操作。对于外连接和上面的执行过程任然适用。例如我们将上面的查询修改如下：
SELECT tbl1.col1 ,tbl2.col2 FROM tbl1 left outer join tbl2 using (col3) WHERE tbl1.col1 in (3,4)
对应的伪代码：
outer_iter = iterator over tbl1 where col1 in(3,4)
outer row = outer_iter.next
while outer_row
inner_iter = iterator over tbl2 where col3 = outer_row.col3
inner_row = inner_iter.next
if inner row -> 手动加粗
while inner_row
out_put [outer_row.col1,inner_row.col2]
inner_row = inner_iter.next
end
else -> 手动加粗
out_put[outer_row.col1,NULL] -> 手动加粗
end
outer_row = outer_iter.next
end
从上面两个例子也可以看出，对于主表来说，是先进行主表的where条件筛选，再进行表联接，而不是先进行整表联接再进行where条件的筛选。
举个例子：
数据表结构：
mysql> create table a(
-> id int unsigned not null primary key
-> );
mysql> create table b like a;
表中数据：
mysql> select * from a;
+----+
| id |
+----+
| 1 |
| 2 |
| 3 |
| 4 |
| 5 |
+----+
5 rows in set (0.00 sec)
mysql> select * from b;
+----+
| id |
+----+
| 4 |
| 5 |
| 6 |
| 7 |
+----+
4 rows in set (0.00 sec)
explain查询：
mysql> explain select a.id as aid,b.id as bid from a left join b using(id) where a.id>3;
+----+-------------+-------+--------+---------------+---------+---------+----------+------+--------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+--------+---------------+---------+---------+----------+------+--------------------------+
| 1 | SIMPLE | a | range | PRIMARY | PRIMARY | 4 | NULL | 3 | Using where; Using index |
| 1 | SIMPLE | b | eq_ref | PRIMARY | PRIMARY | 4 | com.a.id | 1 | Using index |
+----+-------------+-------+--------+---------------+---------+---------+----------+------+--------------------------+
2 rows in set (0.00 sec)
可以看出，首先在a表上进行范围查询，筛选出a.id>3的数据，然后在进行"嵌套查询"。
注意，on后面的筛选条件主要是针对的是关联表，而对于主表筛选并不适用，比如：
mysql> select a.id as aid,b.id as bid from a left join b on a.id=b.id and a.id>3;
+-----+------+
| aid | bid |
+-----+------+
| 1 | NULL |
| 2 | NULL |
| 3 | NULL |
| 4 | 4 |
| 5 | 5 |
+-----+------+
5 rows in set (0.00 sec)
mysql> explain select a.id as aid,b.id as bid from a left join b on a.id=b.id and a.id>3;
+----+-------------+-------+--------+---------------+---------+---------+----------+------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+--------+---------------+---------+---------+----------+------+-------------+
| 1 | SIMPLE | a | index | NULL | PRIMARY | 4 | NULL | 5 | Using index |
| 1 | SIMPLE | b | eq_ref | PRIMARY | PRIMARY | 4 | com.a.id | 1 | Using index |
+----+-------------+-------+--------+---------------+---------+---------+----------+------+-------------+
2 rows in set (0.00 sec)
我们发现a表的id<=3的数据并未被筛选走，explain的结果是a表进行了index类型的查询，即主键索引的全部扫描。
如果在on的筛选条件是针对b表的呢，情况会怎么样？
下面的例子数据表结构和数据变了，我们只关注查询的结果区别：
mysql> select * from a left join b on a.data=b.data and b.id<=20;
+----+------+------+------+
| id | data | id | data |
+----+------+------+------+
| 1 | 1 | NULL | NULL |
| 2 | 2 | NULL | NULL |
| 3 | 3 | NULL | NULL |
| 4 | 4 | 1 | 4 |
| 4 | 4 | 6 | 4 |
| 4 | 4 | 11 | 4 |
| 4 | 4 | 16 | 4 |
| 5 | 5 | 2 | 5 |
| 5 | 5 | 7 | 5 |
| 5 | 5 | 12 | 5 |
| 5 | 5 | 17 | 5 |
+----+------+------+------+
11 rows in set (0.00 sec)
mysql> select * from a left join b on a.data=b.data where b.id<=20;
+----+------+------+------+
| id | data | id | data |
+----+------+------+------+
| 4 | 4 | 1 | 4 |
| 5 | 5 | 2 | 5 |
| 4 | 4 | 6 | 4 |
| 5 | 5 | 7 | 5 |
| 4 | 4 | 11 | 4 |
| 5 | 5 | 12 | 5 |
| 4 | 4 | 16 | 4 |
| 5 | 5 | 17 | 5 |
+----+------+------+------+
8 rows in set (0.00 sec)
由此，我们可以根据伪码来分析两者的区别：
outer_iter = iterator over a
outer row = outer_iter.next
while outer_row
inner_iter = iterator over b where data = outer_row.date where id<=20
inner_row = inner_iter.next
if inner row
while inner_row
out_put [outer_row,inner_row]
inner_row = inner_iter.next
end
else
out_put[outer_row,NULL]
end
outer_row = outer_iter.next
end
outer_iter = iterator over a
outer row = outer_iter.next
while outer_row
inner_iter = iterator over b where data = outer_row.date ->手动加粗
inner_row = inner_iter.next
if inner row
while inner_row
out_put [outer_row,inner_row]
inner_row = inner_iter.next
end
else
out_put[outer_row,NULL]
end
outer_row = outer_iter.next
end
left join的结果集中 where b.id<=20 ->手动加粗
参考资料：《高性能MySQL》