使用 ORDER BY 和 MATCH AGAINST 优化选择查询
Posted
技术标签:
【中文标题】使用 ORDER BY 和 MATCH AGAINST 优化选择查询【英文标题】:optimize select query with ORDER BY and MATCH AGAINST 【发布时间】:2011-05-06 01:37:57 【问题描述】:以下查询需要 25 秒,需要 2 个表连接。第一个帖子表有 150,00 行,主题表有 50,000 行。任何人都知道如何加快速度。
SELECT SQL_NO_CACHE
post_search.post_id,
topic_search.topic_id,
topic_search.topic_title,
topic_search.topic_last_post_time,
MATCH(post_search.post_text,topic_search.topic_title) AGAINST('search_terms' IN BOOLEAN MODE) AS score
FROM bb_posts_fulltext_search post_search
LEFT JOIN bb_topics_fulltext_search topic_search
ON post_search.topic_id = topic_search.topic_id
WHERE MATCH(post_search.post_text,topic_search.topic_title) AGAINST('search_terms' IN BOOLEAN MODE)
GROUP BY topic_search.topic_id
ORDER BY score DESC
LIMIT 0,6
描述
mysql> DESCRIBE bb_posts_fulltext_search;
+-----------+------------+------+-----+---------+-------+
| Field | Type | Null | Key | Default | Extra |
+-----------+------------+------+-----+---------+-------+
| post_id | bigint(20) | NO | PRI | NULL | |
| post_text | longtext | YES | MUL | NULL | |
| topic_id | bigint(20) | YES | MUL | NULL | |
+-----------+------------+------+-----+---------+-------+
描述
mysql> DESCRIBE bb_topics_fulltext_search
-> ;
+----------------------+--------------+------+-----+---------+-------+
| Field | Type | Null | Key | Default | Extra |
+----------------------+--------------+------+-----+---------+-------+
| topic_id | int(11) | NO | PRI | NULL | |
| topic_title | varchar(255) | YES | MUL | NULL | |
| topic_posts | bigint(20) | YES | | NULL | |
| topic_poster_name | varchar(40) | YES | | NULL | |
| topic_last_post_id | bigint(20) | YES | | NULL | |
| forum_id | int(11) | YES | | NULL | |
| parent_group_id | int(11) | YES | | NULL | |
| child_group_id | int(11) | YES | | NULL | |
| topic_last_post_time | datetime | YES | MUL | NULL | |
+----------------------+--------------+------+-----+---------+-------+
解释
+----+-------------+--------------+--------+---------------+---------+---------+----------------------------------+--------+---------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+--------------+--------+---------------+---------+---------+----------------------------------+--------+---------------------------------+
| 1 | SIMPLE | post_search | ALL | NULL | NULL | NULL | NULL | 158972 | Using temporary; Using filesort |
| 1 | SIMPLE | topic_search | eq_ref | PRIMARY | PRIMARY | 4 | wordpress.post_search.topic_id | 1 | Using where |
+----+-------------+--------------+--------+---------------+---------+---------+----------------------------------+--------+---------------------------------+
更新:索引
+--------------------------+------------+-----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment |
+--------------------------+------------+-----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| bb_posts_fulltext_search | 0 | PRIMARY | 1 | post_id | A | 158972 | NULL | NULL | | BTREE | | |
| bb_posts_fulltext_search | 1 | topic_id | 1 | topic_id | A | 52990 | NULL | NULL | YES | BTREE | | |
| bb_posts_fulltext_search | 1 | post_text | 1 | post_text | NULL | 1 | NULL | NULL | YES | FULLTEXT | | |
+--------------------------+------------+-----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
创建 1
DROP TABLE IF EXISTS `wordpress`.`bb_posts_fulltext_search`;
CREATE TABLE `wordpress`.`bb_posts_fulltext_search` (
`post_id` bigint(20) NOT NULL,
`post_text` longtext,
`topic_id` bigint(20) DEFAULT NULL,
PRIMARY KEY (`post_id`),
KEY `topic_id` (`topic_id`),
FULLTEXT KEY `post_text` (`post_text`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;
创建 2
DROP TABLE IF EXISTS `wordpress`.`bb_topics_fulltext_search`;
CREATE TABLE `wordpress`.`bb_topics_fulltext_search` (
`topic_id` int(11) NOT NULL,
`topic_title` varchar(255) DEFAULT NULL,
`topic_posts` bigint(20) DEFAULT NULL,
`topic_poster_name` varchar(40) DEFAULT NULL,
`topic_last_post_id` bigint(20) DEFAULT NULL,
`forum_id` int(11) DEFAULT NULL,
`parent_group_id` int(11) DEFAULT NULL,
`child_group_id` int(11) DEFAULT NULL,
`topic_last_post_time` datetime DEFAULT NULL,
PRIMARY KEY (`topic_id`),
KEY `topic_last_post_time` (`topic_last_post_time`),
FULLTEXT KEY `topic_title` (`topic_title`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;
解决方案: http://forums.mysql.com/read.php?115,418955,418955#msg-418955
【问题讨论】:
你用的是左连接,有没有没有主题的帖子? FULLTEXT 索引必须适合您的 myisam_key_buffer,因此请确保正确设置此配置变量。 FULLTEXT 的性能也是出了名的糟糕。使用像 Xapian 这样的真正搜索引擎,我得到了更多可用的结果(比如快 1000 倍)。 Sphing 和 Lucene 也有很好的声誉。 您似乎找到了解决问题的方法。我建议您写一个简单的解释,将其作为答案发布,然后接受它,以便我们记录如何解决问题,因此我们不会悬而未决这个问题。谢谢。 【参考方案1】:OP在这里找到了答案,以防万一链接消失,下面是该链接中的文字:
DROP TABLE IF EXISTS `wordpress`.`bb_posts_fulltext_search`;
CREATE TABLE `wordpress`.`bb_posts_fulltext_search` (
`post_id` int(10) unsigned NOT NULL,
`post_text` longtext,
`topic_id` int(10) unsigned NOT NULL DEFAULT '0',
PRIMARY KEY (`post_id`,`topic_id`) USING BTREE,
FULLTEXT KEY `post_text` (`post_text`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;
DROP TABLE IF EXISTS `wordpress`.`bb_topics_fulltext_search`;
CREATE TABLE `wordpress`.`bb_topics_fulltext_search` (
`topic_id` int(11) NOT NULL,
`topic_title` varchar(255) DEFAULT NULL,
`topic_posts` bigint(20) DEFAULT NULL,
`topic_poster_name` varchar(40) DEFAULT NULL,
`topic_last_post_id` bigint(20) DEFAULT NULL,
`forum_id` int(11) DEFAULT NULL,
`parent_group_id` int(11) DEFAULT NULL,
`child_group_id` int(11) DEFAULT NULL,
`topic_last_post_time` datetime DEFAULT NULL,
PRIMARY KEY (`topic_id`),
KEY `topic_last_post_time` (`topic_last_post_time`),
FULLTEXT KEY `topic_title` (`topic_title`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;
SELECT
topic_search.topic_id,
topic_search.topic_title,
topic_search.topic_posts,
topic_search.topic_title,
topic_search.topic_poster_name,
topic_search.topic_last_post_id,
topic_search.topic_last_post_time,
MATCH(post_search.post_text,topic_search.topic_title)
AGAINST('searchterms' IN BOOLEAN MODE) AS score
FROM bb_posts_fulltext_search as post_search
LEFT JOIN bb_topics_fulltext_search as topic_search
ON post_search.topic_id = topic_search.topic_id
WHERE
MATCH(post_search.post_text,topic_search.topic_title)
AGAINST('searchterms' IN BOOLEAN MODE)
GROUP BY topic_search.topic_id
ORDER BY score DESC
LIMIT 0,6
【讨论】:
以上是关于使用 ORDER BY 和 MATCH AGAINST 优化选择查询的主要内容,如果未能解决你的问题,请参考以下文章