explain select goods_id,cat_id,goods_name from  ecs_goods where cat_id in (select cat_id from ecs_category where parent_id=6) \G:

in子查询陷阱_in

(外面的查询type是 all,也就是整张表都被扫了一遍,若goods表数据够多,那就很慢了;子查询语句倒是很快,rows才1)

如果子查询语句替换成具体的id:

explain select goods_id,goods_name from  ecs_goods where cat_id in (1,2,3,4,5,6,7,8,9,11,12,13,14,15) \G

in子查询陷阱_查询_02

而,少几个cat_id:explain select goods_id,goods_name from  ecs_goods where cat_id in (7,8,9,11) \G:

in子查询陷阱_查询_03

而如果用上join语句:explain select goods_id,goods_name from ecs_goods inner join (select cat_id from ecs_category) as tmp on tmp.cat_id=ecs_goods.cat_id \G:

in子查询陷阱_陷阱_04

(这时候不会再扫描31行了,而是14行,等于category中的条数)




: ecshop商城表中,查询6号栏目的商品, (,6号是一个大栏目)

最直观的: mysql> select goods_id,cat_id,goods_name from  goods where cat_id in (select

cat_id from ecs_category where parent_id=6);

误区: 给我们的感觉是, 先查到内层的6号栏目的子栏目,7,8,9,11

然后外层, cat_id in (7,8,9,11)

 

事实: 如下图, goods表全扫描, 并逐行与category表对照,parent_id=6是否成立

in子查询陷阱_in_05

原因: mysql的查询优化器,针对In型做优化,被改成了exists的执行效果.

goods表越大时, 查询速度越慢.

 

改进: 用连接查询来代替子查询

 explain select goods_id,g.cat_id,g.goods_name from  goods as g

 inner join (select cat_id from ecs_category where parent_id=6) as t

 using(cat_id) \G

 

内层 select cat_id from ecs_category where parent_id=6 ; 用到Parent_id索引, 返回4

+--------+

| cat_id |

+--------+

|      7 |

|      8 |

|      9 |

|     11 |

+--------+    形成结果,设为t


*************************** 3. row ***************************

           id: 2

  select_type: DERIVED

        table: ecs_category

         type: ref

possible_keys: parent_id

          key: parent_id

      key_len: 2

          ref:

         rows: 4

        Extra:

3 rows in set (0.00 sec)

 

 

2次查询,

tgoods 通过 cat_id 相连,

因为cat_idgoods表中有索引, 所以相当于用7,8,911,快速匹配上 goods的行.

*************************** 2. row ***************************

           id: 1

  select_type: PRIMARY

        table: g

         type: ref

possible_keys: cat_id

          key: cat_id

      key_len: 2

          ref: t.cat_id

         rows: 6

        Extra:

 

1次查询 :

是把上面2次的中间结果,直接取回.

*************************** 1. row ***************************

           id: 1

  select_type: PRIMARY

        table: <derived2>

         type: ALL

possible_keys: NULL

          key: NULL

      key_len: NULL

          ref: NULL

         rows: 4

        Extra:


所以,不能简单的说join语句速度就慢,要想知道具体情况下的性能,就要explain