使用LEFT JOIN进行MySQL查询优化

Question

我有一个优化的MySQL查询，因为运行此查询时，它将MySQL进程的CPU使用率提高到100％。 因此，我对该查询做了一些修改，现在它减少了其运行时间，但CPU使用率仍接近100％。 我在两个表中都索引了所有搜索字段。

这些“ milestone_and_events”和“ milestone_and_events_reminders”具有一对多的关系（“ milestone_and_events”具有一个或多个“ milestone_and_events_reminders”），并且“ milestone_and_events_reminders”表具有近150万条记录。

有什么办法可以进一步优化此查询，任何人都可以给我一个指南吗？ 谢谢

这是原始查询

SELECT MAE.*
     , MAER.reminder_user_id
     , MAER.reminder_type
     , MAER.id as reminder_id 
  FROM milestone_and_events AS MAE 
  LEFT 
  JOIN milestone_and_events_reminders MAER 
    ON MAE.id = MAER.milestone_and_events_id 
 WHERE MAER.alert_time < '$currentTime'
   AND MAER.issued ! = 1 
   AND MAE.completed = 0
 GROUP 
    BY MAE.id
     , MAER.reminder_user_id
     , MAER.reminder_type

这是我目前的查询

$currentTime = date('Y-m-d H:i:s');
$query = 'SELECT MAE.id, MAE.time, MAER.reminder_user_id, MAER.reminder_type, MAER.id AS reminder_id 
            FROM milestone_and_events AS MAE 
            LEFT JOIN milestone_and_events_reminders MAER ON 
            MAE.id = MAER.milestone_and_events_id AND 
            MAER.issued =0 AND 
            MAER.alert_time < "' . $currentTime . '" AND 
            MAE.completed =0 
            WHERE MAER.reminder_type != "onTime"
            GROUP BY MAER.milestone_and_events_id,MAER.reminder_user_id,MAER.reminder_type';

更新1

此“ milestone_and_events”表具有近200个条目。

CREATE TABLE `milestone_and_events` (
 `id` int(11) NOT NULL AUTO_INCREMENT,
 `description` varchar(250) NOT NULL, 
 `time` datetime NOT NULL,
 `attached_type` int(11) DEFAULT NULL,
 `attached_type_value` int(11) DEFAULT NULL, 
 `completed` int(11) NOT NULL DEFAULT '0', 
 `type` int(11) NOT NULL DEFAULT '1', 
 PRIMARY KEY (`id`),
 KEY `time` (`time`), 
 KEY `completed` (`completed`),
 KEY `id` (`id`,`completed`)
) ENGINE=InnoDB AUTO_INCREMENT=154 DEFAULT CHARSET=utf8

“ milestone_and_events_reminders”表具有近150万个条目。

CREATE TABLE `milestone_and_events_reminders` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `reminder_user_id` int(11) NOT NULL,
  `milestone_and_events_id` int(11) NOT NULL,
  `alert_time` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
  `issued` tinyint(11) NOT NULL DEFAULT '0',
  `reminder_type` enum('upComming','delayed','onTime') NOT NULL,
  PRIMARY KEY (`id`),
  KEY `user_id` (`reminder_user_id`),
  KEY `reminder_type` (`reminder_type`),
  KEY `issued` (`issued`),
  KEY `milestone_and_events_id` (`milestone_and_events_id`, `issued`,`reminder_type`),
   CONSTRAINT `milestone_and_events_reminders_ibfk_1` FOREIGN KEY                       (`milestone_and_events_id`) 
   REFERENCES `milestone_and_events` (`id`) ON DELETE NO ACTION ON UPDATE NO ACTION
   ) ENGINE=InnoDB AUTO_INCREMENT=3323544 DEFAULT CHARSET=utf8

Answer 1

这是您的查询：

SELECT MAE.id, MAE.time, MAER.reminder_user_id, MAER.reminder_type, MAER.id AS reminder_id 
FROM milestone_and_events AS MAE LEFT JOIN
     milestone_and_events_reminders MAER
     ON MAE.id = MAER.milestone_and_events_id AND 
        MAER.issued = 0 AND 
        MAER.alert_time < "' . $currentTime . '"            
WHERE MAE.completed = 0 AND MAER.reminder_type <> "onTime"
GROUP BY MAE.id, MAER.reminder_user_id, MAER.reminder_type;

注意：我将条件MAE.completed = 0从on子句移到了where子句。 它在on子句中什么也不做。 而且，我更改了group by的第一个键以匹配SELECT 。

此查询的最佳索引是复合索引： milestone_and_events(completed, id)和milestone_and_events_reminders(milestone_and_events_id, issued, alert_time, reminder_type) 。

我的猜测是，将completed = 0放在其所属的位置将减少数据量并提高查询的性能。