如何通过子查询操作优化计数和分组答案

【问题标题】：How to optimize count and group by subquery operation如何通过子查询操作优化计数和分组
【发布时间】：2013-09-06 21:49:51
【问题描述】：

我的主机说以下查询占用了大量服务器 CPU。请告诉我如何优化它。

SELECT COUNT(*) FROM (SELECT COUNT(*) AS tot,wallpapers.*,resolutions.res_height,resolutions.res_width FROM wallpapers

INNER JOIN analytics ON analytics.`wall_id` = wallpapers.`wall_id`

INNER JOIN resolutions ON resolutions.`res_id` = wallpapers.`res_id`



WHERE analytics.ana_date >= '2013-09-01 16:36:56' AND wallpapers.wall_status = 'public'

GROUP BY analytics.`wall_id`) as Q

请注意，分析表包含所有网页浏览和点击的记录。所以它非常非常大。

【问题讨论】：

标签： mysql sql database optimization count

【解决方案1】：

据我所知，在通过连接和 WHERE 子句过滤后，您的查询只计算不同的 wall_id 值。这样的事情应该很接近：

SELECT COUNT(DISTINCT analytics.wall_id)
FROM wallpapers
INNER JOIN analytics ON analytics.wall_id = wallpapers.wall_id
INNER JOIN resolutions ON resolutions.res_id = wallpapers.res_id
WHERE analytics.ana_date >= '2013-09-01 16:36:56'
  AND wallpapers.wall_status = 'public'

【讨论】：

【解决方案2】：

这是您的查询：

SELECT COUNT(*)
FROM (SELECT COUNT(*) AS tot, wallpapers.*, resolutions.res_height, resolutions.res_width
      FROM wallpapers INNER JOIN
           analytics
           ON analytics.`wall_id` = wallpapers.`wall_id` INNER JOIN
           resolutions
           ON resolutions.`res_id` = wallpapers.`res_id`
      WHERE analytics.ana_date >= '2013-09-01 16:36:56' AND
            wallpapers.wall_status = 'public'
      GROUP BY analytics.`wall_id`
) as Q

子查询需要额外的努力，group by 也是如此。您可以将其替换为：

SELECT COUNT(distinct analytics.wall_id)
FROM wallpapers INNER JOIN
     analytics
     ON analytics.`wall_id` = wallpapers.`wall_id` INNER JOIN
     resolutions
     ON resolutions.`res_id` = wallpapers.`res_id`
WHERE analytics.ana_date >= '2013-09-01 16:36:56' AND
      wallpapers.wall_status = 'public';

然后您也许可以使用索引进行进一步优化，但查看此查询的 explain 和表上的当前索引会很有帮助。

【讨论】：