0

我在 MySQL 表中查询为特定 x/y/z 坐标记录的事件（对于游戏插件，仅供参考）。此查询是使用右连接编写的，以排除在完全相同坐标处具有较新条目的记录。

然而，这是一个性能杀手。

例如，一个查询返回大约 32k 个结果（它们通常不会这么高，但我需要解决根本问题） -如果没有正确的连接比较，它需要：33373 行集合（0.19 秒）。

使用正确的连接，它需要 32955 行集合（3 分 7.99 秒）

3分钟太长了。

这是一个包含右连接的示例查询：

SELECT prism_actions.id, prism_actions.action_time, prism_actions.action_type, prism_actions.player, prism_actions.world, prism_actions.x, prism_actions.y, prism_actions.z, prism_actions.data, DATE_FORMAT(prism_actions.action_time, '%c/%e/%y') display_date, DATE_FORMAT(prism_actions.action_time, '%l:%i%p') display_time
FROM prism_actions
RIGHT JOIN (SELECT action_type, x, y, z, max(action_time) as action_time FROM prism_actions GROUP BY action_type, x, y, z) latest 
ON prism_actions.action_time = latest.action_time
AND prism_actions.x = latest.x
AND prism_actions.y = latest.y
AND prism_actions.z = latest.z
AND prism_actions.action_type = latest.action_type
WHERE world = 'world'
AND (prism_actions.action_type = 'creeper-explode' OR prism_actions.action_type = 'entity-explode' OR prism_actions.action_type = 'tnt-explode' OR prism_actions.action_type = 'block-burn')
AND LEFT(prism_actions.action_type,5) != 'prism'
AND (prism_actions.x BETWEEN 2412.0286793077976 AND 2612.0286793077976)
AND (prism_actions.y BETWEEN -25.5 AND 174.5)
AND (prism_actions.z BETWEEN -2650.697295131335 AND -2450.697295131335)
ORDER BY prism_actions.action_time ASC, x ASC, z ASC, y ASC, id ASC
LIMIT 0,1000000

没有，就是简单地删除正确的连接。

所以我的问题是：

  • 这是处理排除具有较新条目的记录的最佳方法吗？
  • 如果是这样，我可以做些什么来提高性能吗？

我使用这样的查询的原因是我的软件需要按时间戳排序的结果，但应该始终只知道每个 x/y/z 的最新活动，否则会发生冲突。

表结构：

CREATE TABLE IF NOT EXISTS `prism_actions` (
  `id` int(11) unsigned NOT NULL auto_increment,
  `action_time` datetime NOT NULL,
  `action_type` varchar(25) NOT NULL,
  `player` varchar(16) NOT NULL,
  `world` varchar(255) NOT NULL,
  `x` int(11) NOT NULL,
  `y` int(11) NOT NULL,
  `z` int(11) NOT NULL,
  `data` varchar(255) NOT NULL,
  PRIMARY KEY  (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=latin1;

解释上述查询的非常减少的区域的结果：

+----+-------------+---------------+------+---------------+------+---------+------+------+----------------------------------------------+
| id | select_type | table         | type | possible_keys | key  | key_len | ref  | rows | Extra                                        |
+----+-------------+---------------+------+---------------+------+---------+------+------+----------------------------------------------+
|  1 | PRIMARY     | <derived2>    | ALL  | NULL          | NULL | NULL    | NULL | 1870 | Using where; Using temporary; Using filesort | 
|  1 | PRIMARY     | prism_actions | ALL  | NULL          | NULL | NULL    | NULL | 1970 | Using where                                  | 
|  2 | DERIVED     | prism_actions | ALL  | NULL          | NULL | NULL    | NULL | 1970 | Using temporary; Using filesort              | 
+----+-------------+---------------+------+---------------+------+---------+------+------+----------------------------------------------+
3 rows in set (0.04 sec)
4

1 回答 1

0

我的猜测是下面的工作会稍微好一些。假设它为您提供了正确的答案，但对性能没有太大帮助，请考虑为相关表提供 SHOW CREATE TABLE 语句，并在此查询上发布 EXPLAIN 的输出......

SELECT pa.id
     , pa.action_time
     , pa.action_type
     , pa.player
     , pa.world
     , pa.x
     , pa.y
     , pa.z
     , pa.data
     , DATE_FORMAT(pa.action_time, '%c/%e/%y') display_date
     , DATE_FORMAT(pa.action_time, '%l:%i%p') display_time
  FROM prism_actions pa 
  JOIN 
     ( SELECT action_type
            , x
            , y
            , z
            , MAX(action_time) max_action_time 
         FROM prism_actions 
        GROUP 
           BY action_type
            , x
            , y
            , z
     ) latest 
    ON latest.max_action_time = pa.action_time 
   AND pa.x = latest.x
   AND pa.y = latest.y
   AND pa.z = latest.z
   AND pa.action_type = latest.action_type
 WHERE world = 'world'
   AND pa.action_type IN('creeper-explode','entity-explode','tnt-explode','block-burn')
   AND pa.action_type NOT LIKE 'prism%'
   AND pa.x BETWEEN 2412.0286793077976 AND 2612.0286793077976
   AND pa.y BETWEEN -25.5 AND 174.5
   AND pa.z BETWEEN -2650.697295131335 AND -2450.697295131335
 ORDER 
    BY pa.action_time ASC
     , x ASC
     , z ASC
     , y ASC
     , id ASC
 LIMIT 0,1000000;
于 2013-01-12T18:35:27.337 回答