Question

今天我收到了来自我的托管帐户的电子邮件，说我需要调整我的查询：

SELECT
  `id`, `nick`, `msg`, `uid`, `show_pic`,
  `time`,`ip`,`time_updated`,
  (SELECT COUNT(c.msg_id)
   FROM `the_ans` c
   where c.msg_id = d.id) AS counter,
  (SELECT c.msg
   FROM `the_ans` c
   WHERE c.msg_id=d.id
   ORDER BY `time` DESC LIMIT 1) as lastmsg
FROM
  `the_data` d
ORDER BY `time_updated` DESC LIMIT 26340 ,15

说明：

id select_type table type possible_keys key key_len ref rows Extra
1 PRIMARY d ALL 34309 Using filesort
3 DEPENDENT SUBQUERY c ALL 43659 Using where; Using filesort
2 DEPENDENT SUBQUERY c ALL 43659 Using where

此查询检查65,396,669,012,829行，这在共享主机中是不可接受的。

tbh，我不明白他们的解释.. 查询实际上做的是按时间顺序更新15个帖子，对于每个帖子，我抓住最新的评论，计算每篇文章的所有评论。

posts table - 'the_data'

comments table = 'the_ans'

我不是一个mysql大师，我不知道如何改进这个查询任何帮助将不胜感激

THX

查询

SELECT
  `id` , `nick` , `msg` , `uid` , `show_pic` , `time` , `ip` , `time_updated` , (
    SELECT COUNT( c.msg_id )
    FROM `the_ans` c
    WHERE c.msg_id = d.id
   ) AS counter, (
    SELECT c.msg
    FROM `the_ans` c
    WHERE c.msg_id = d.id
    ORDER BY `time` DESC
    LIMIT 1
   ) AS lastmsg
FROM `the_data` d
ORDER BY `time_updated` DESC
LIMIT 26340 , 15

这是结果结构

id| nick  | msg  | uid   | show_pick | time      | ip |time_updated|counter|lastmsg
  |       |      |       |           |           |    |            |       |
7 | jqman | hello| 10074 |   0       |2013-21-01 | 12 |2013-21-01  | 55    |blah bl

Answer 1

快速浏览解释计划表明，MySQL没有合适的索引可供使用，因此它采用全表扫描。

 EXPLAIN: 
 id select_type        table type possible_keys key key_len ref rows  Extra 
 -- ------------------ ----- ---- ------------- --- ------- --- ----- ---------------------------- 
 1  PRIMARY            d     ALL                                34309 Using filesort
 3  DEPENDENT SUBQUERY c     ALL                                43659 Using where; Using filesort 
 2  DEPENDENT SUBQUERY c     ALL                                43659 Using where

要优化现有查询的执行，需要添加适当的索引。可能的候选人：

ON `the_data`(`time_updated`)
ON `the_ans`(`msg_id`,`time`)

这些索引将显着提高外部查询（可能消除排序操作）的性能，以及相关子查询的大量执行。

除此之外，您还需要更改查询以提高性能。在准备整个结果集之后，将应用最外层查询的LIMIT子句，这意味着将为表the_data中的每一行执行这两个相关子查询。这将是你的午餐，表现明智。

要使这些相关子查询仅针对返回的（最多）15行运行，您需要在这些子查询运行之前应用LIMIT子句。

此查询应返回等效的结果集，并且将避免每个相关子查询的34,000多次执行，这应该会显着提高性能：

SELECT d.*
     , ( SELECT COUNT( c.msg_id )
           FROM `the_ans` c
          WHERE c.msg_id = d.id
       ) AS counter
     , ( SELECT c.msg
           FROM `the_ans` c
          WHERE c.msg_id = d.id
          ORDER BY `time` DESC
          LIMIT 1
       ) AS lastmsg
  FROM ( SELECT e.`id` 
              , e.`nick`
              , e.`msg`
              , e.`uid`
              , e.`show_pic`
              , e.`time`
              , e.`ip`
              , e.`time_updated` 
           FROM `the_data` e
          ORDER
             BY e.`time_updated` DESC
          LIMIT 26340 , 15 
       ) d
 ORDER BY d.`time_updated` DESC

（您当前的查询执行每个相关子查询“SELECT COUNT(1) FROM the_data”次。通过上面的重写查询，每个子查询只执行15次。）

Answer 2

在从主查询中选择限时行后执行相关子查询：

SELECT d.*, (SELECT COUNT(c.msg_id) FROM `the_ans` c where c.msg_id = d.id) AS counter, (SELECT c.msg FROM `the_ans` c WHERE c.msg_id=d.id ORDER BY `time` DESC LIMIT 1) as lastmsg FROM (SELECT `id`, `nick`, `msg`, `uid`, `show_pic`, `time`,`ip`,`time_updated` FROM `the_data` ORDER BY `time_updated` DESC LIMIT 26340 ,15) d

另外，请确保您在time_updated和msg_id上有索引。

Answer 3

这样的事情会让你的结果更快一些。

请注意，这是使用INNER JOIN，因为当* the_data *上的每条记录在* the_ans *

上至少有一条匹配记录时，它可以正常工作

SELECT `id` , `nick` , `msg` , `uid` , `show_pic` , `time` , `ip` , `time_updated` , Sub1.counter, c.msg AS lastmsg
FROM `the_data` d
INNER JOIN (SELECT msg_id, COUNT( * ) AS counter, MAX( `time` ) AS MaxTime FROM `the_ans` GROUP BY msg_id) Sub1 ON d.id = Sub1.msg_id
INNER JOIN the_ans c ON d.id = c.msg_id AND sub1.MaxTime = c.`time`
ORDER BY `time_updated` DESC
LIMIT 26340 , 15

带子查询的MySQL优化查询

3 个答案: