如何改进MySQL中的Limit子句

时间:2015-04-15 11:34:11

标签: mysql sql performance limit

我有{10}行的posts表,我希望通过它创建分页。所以我有下一个查询:

SELECT post_id
    FROM posts
    LIMIT 0, 10;

当我Explain查询时,我得到了下一个结果:

enter image description here

所以我不明白为什么MySql需要通过9976行迭代才能找到10个第一行?如果有人帮我优化这个查询,我将非常感激。

我也知道该主题MySQL ORDER BY / LIMIT performance: late row lookups,但即使我将查询修改为下一个问题,问题仍然存在:

SELECT  t.post_id
FROM    (
        SELECT  post_id
        FROM    posts
        ORDER BY
                post_id
        LIMIT 0, 10
        ) q 
JOIN    posts t 
ON      q.post_id = t.post_id

enter image description here

更新

@ pala_的解决方案适用于上述简单案例,但现在我正在使用inner join测试更复杂的查询。我的目的是将评论表与帖子表联系起来,不幸的是,当我解释新查询仍在迭代9976行时。

Select comm.comment_id 
from comments as comm 
    inner join (
        SELECT post_id 
        FROM posts 
        ORDER BY post_id 
        LIMIT 0, 10
    ) as paged_post on comm.post_id = paged_post.post_id;  

您是否知道这种MySQL行为的原因是什么?

1 个答案:

答案 0 :(得分:5)

试试这个:

SELECT post_id
    FROM posts
    ORDER BY post_id DESC
    LIMIT 0, 10;

通过LIMIT的分页在没有订购的情况下没有多大意义,它应该可以解决您的问题。

mysql> explain select * from foo;
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table | type  | possible_keys | key     | key_len | ref  | rows | Extra       |
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
|  1 | SIMPLE      | foo   | index | NULL          | PRIMARY | 4       | NULL |   20 | Using index |
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
1 row in set (0.00 sec)

mysql> explain select * from foo limit 0, 10;
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table | type  | possible_keys | key     | key_len | ref  | rows | Extra       |
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
|  1 | SIMPLE      | foo   | index | NULL          | PRIMARY | 4       | NULL |   20 | Using index |
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
1 row in set (0.00 sec)

mysql> explain select * from foo order by id desc limit 0, 10;
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table | type  | possible_keys | key     | key_len | ref  | rows | Extra       |
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
|  1 | SIMPLE      | foo   | index | NULL          | PRIMARY | 4       | NULL |   10 | Using index |
+----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
1 row in set (0.00 sec)

关于您对评论加入的最新评论。你有comment(post_id)的索引吗?根据我的测试数据,我得到以下结果:

mysql> alter table comments add index pi (post_id);
Query OK, 0 rows affected (0.15 sec)
Records: 0  Duplicates: 0  Warnings: 0

mysql> explain select c.id from  comments c inner join (select id from posts o order by id  limit 0, 10) p on c.post_id = p.id;
+----+-------------+------------+-------+---------------+---------+---------+------+------+--------------------------+
| id | select_type | table      | type  | possible_keys | key     | key_len | ref  | rows | Extra                    |
+----+-------------+------------+-------+---------------+---------+---------+------+------+--------------------------+
|  1 | PRIMARY     | <derived2> | ALL   | NULL          | NULL    | NULL    | NULL |   10 |                          |
|  1 | PRIMARY     | c          | ref   | pi            | pi      | 5       | p.id |    4 | Using where; Using index |
|  2 | DERIVED     | o          | index | NULL          | PRIMARY | 4       | NULL |   10 | Using index              |
+----+-------------+------------+-------+---------------+---------+---------+------+------+--------------------------+

和表格大小参考:

mysql> select count(*) from posts;
+----------+
| count(*) |
+----------+
|    15021 |
+----------+
1 row in set (0.01 sec)

mysql> select count(*) from comments;
+----------+
| count(*) |
+----------+
|     1000 |
+----------+
1 row in set (0.00 sec)