Question

我在MySql（5.1）InnoDB中有一个查询，它在一个包含部分的表中搜索。带有零件的表包含大约500 000行。搜索还加入了另外两个表tblcategory和tblheadcategory。我有很多用户使用这个查询，它使我的服务器几乎崩溃负载很重。

我知道一个好方法是使用全文搜索，我希望我们可以改变它以便将来使用它。但是因为InnoDB无法做到这一点，我需要一个快速的＆＃34;优化让它现在运行。我应该如何优化这个并设置索引和其他东西，以使此查询尽可能好地运行？

这是查询：

SELECT tblpart.partid,tblpart.title,tblcategory.category,tblheadcategory.headcategory

FROM tblpart

INNER JOIN tblcategory ON tblpart.categoryid = tblcategory.categoryid
INNER JOIN tblheadcategory ON tblcategory.headcategoryid = tblheadcategory.headcategoryid

WHERE (tblpart.title LIKE '%bmw%' OR tblpart.description LIKE '%bmw%' OR tblpart.brand LIKE '%bmw%')

ORDER BY

tblpart.title='bmw' DESC,
tblcategory.category LIKE '%bmw%' DESC

LIMIT 50;

表格：

CREATE TABLE `tblpart` (
    `partid` int(10) NOT NULL auto_increment,
    `userid` int(11) default '1',
    `categoryid` int(10) default '1',
    `title` varchar(100) default NULL,
    `brand` varchar(100) default NULL,
    `description` varchar(100) default NULL,
    PRIMARY KEY  (`partid`),
    KEY `userid` (`userid`),
    KEY `title` (`title`)
) ENGINE=InnoDB AUTO_INCREMENT=534007 DEFAULT CHARSET=utf8;

CREATE TABLE `tblcategory` (
    `categoryid` int(10) NOT NULL auto_increment,
    `category` varchar(255) default NULL,
    `headcategoryid` int(10) default NULL,
      PRIMARY KEY  (`categoryid`)
) ENGINE=InnoDB AUTO_INCREMENT=1261 DEFAULT CHARSET=utf8;

CREATE TABLE `tblheadcategory` (
    `headcategoryid` int(10) NOT NULL auto_increment,
    `headcategory` varchar(255) default NULL,
    PRIMARY KEY  (`headcategoryid`)
) ENGINE=InnoDB AUTO_INCREMENT=21 DEFAULT CHARSET=utf8;

EXPLAIN提供以下内容:(抱歉，我无法弄清楚如何正确格式化）

id   select_type   table            type    possible_keys   key      key_len  ref                         rows        extra
1    SIMPLE        tblpart          ALL     NULL            NULL     NULL     NULL                        522905      Using where; Using temporary; Using filesort
1    SIMPLE        tblcategory      eq_ref  PRIMARY         PRIMARY  4        tblpart.categoryid          1
1    SIMPLE        tblheadcategory  eq_ref  PRIMARY         PRIMARY  4        tblcategory.headcategoryid  1

更新

根据我尝试FULLTEXT解决方案的建议：

新的MyISAM表：

CREATE TABLE `tblpart_search` (
    `partid` int(11) NOT NULL AUTO_INCREMENT,
    `title` varchar(100) NOT NULL,
    `brand` varchar(100) DEFAULT NULL,
    `description` varchar(100) DEFAULT NULL,
    PRIMARY KEY (`partid`),
    FULLTEXT KEY `all` (`title`,`brand`,`description`)
) ENGINE=MyISAM AUTO_INCREMENT=359596 DEFAULT CHARSET=utf8;

触发器：

DELIMITER ;;
CREATE TRIGGER `tblpart_insert_trigger` AFTER INSERT ON `tblpart` 
FOR EACH ROW INSERT INTO tblpart_search VALUES(NEW.partid,NEW.title,NEW.brand,NEW.description);;
DELIMITER ;

DELIMITER ;;
CREATE TRIGGER `tblpart_update_trigger` AFTER UPDATE ON `tblpart` 
FOR EACH ROW UPDATE tblpart_search SET tblpart_search.title=NEW.title,tblpart_search.brand=NEW.brand,tblpart_search.description=NEW.description WHERE tblpart_search.partid=NEW.partid;;
DELIMITER ;

DELIMITER ;;
CREATE TRIGGER `tblpart_delete_trigger` AFTER DELETE ON `tblpart` 
FOR EACH ROW DELETE FROM tblpart_search WHERE tblpart_search.partid=OLD.partid;;
DELIMITER ;

新查询：

SELECT tblpart.partid,tblpart.title,tblcategory.category,tblheadcategory.headcategory

FROM tblpart_search
INNER JOIN tblpart ON tblpart_search.partid = tblpart.partid
INNER JOIN tblcategory ON tblpart.categoryid = tblcategory.categoryid
INNER JOIN tblheadcategory ON tblcategory.headcategoryid = tblheadcategory.headcategoryid

WHERE MATCH (tblpart_search.title, tblpart_search.brand, tblpart_search.description) AGAINST ('bmw,car')
LIMIT 50;

Answer 1

您无法使用前导通配符真正优化查询（即使使用FULLTEXT次搜索）。

这里唯一可以做的就是将查询拆分为三个（在客户端）：

SELECT  tblpart.partid,tblpart.title,tblcategory.category,tblheadcategory.headcategory
FROM    tblpart
INNER JOIN
        tblcategory
ON      tblpart.categoryid = tblcategory.categoryid
INNER JOIN
        tblheadcategory
ON      tblcategory.headcategoryid = tblheadcategory.headcategoryid
WHERE   tblpart.title = 'bmw'
ORDER BY
        tblcategory.category LIKE '%bmw%' DESC
LIMIT 50

SELECT  tblpart.partid,tblpart.title,tblcategory.category,tblheadcategory.headcategory
FROM    tblpart
INNER JOIN
        tblcategory
ON      tblpart.categoryid = tblcategory.categoryid
INNER JOIN
        tblheadcategory
ON      tblcategory.headcategoryid = tblheadcategory.headcategoryid
WHERE   tblpart.title <> 'bmw'
        AND  (tblpart.title LIKE '%bmw%' OR tblpart.description LIKE '%bmw%' OR tblpart.brand LIKE '%bmw%')
        AND tblcategory.category LIKE '%bmw%'
LIMIT N

SELECT  tblpart.partid,tblpart.title,tblcategory.category,tblheadcategory.headcategory
FROM    tblpart
INNER JOIN
        tblcategory
ON      tblpart.categoryid = tblcategory.categoryid
INNER JOIN
        tblheadcategory
ON      tblcategory.headcategoryid = tblheadcategory.headcategoryid
WHERE   tblpart.title <> 'bmw'
        AND  (tblpart.title LIKE '%bmw%' OR tblpart.description LIKE '%bmw%' OR tblpart.brand LIKE '%bmw%')
        AND tblcategory.category NOT LIKE '%bmw%'
LIMIT N

并使用N替换上次查询中的50 - records，其中records是先前查询返回的记录数

第一个查询可以使用title上的索引提供。

<强>更新

FULLTEXT搜索可以像这样实现：

CREATE TABLE `tblpart_search` (
    `partid` int(11) NOT NULL AUTO_INCREMENT,
    `title` varchar(100) NOT NULL,
    `brand` varchar(100) DEFAULT NULL,
    `description` varchar(100) DEFAULT NULL,
    PRIMARY KEY (`partid`),
    FULLTEXT KEY `all` (`title`,`brand`,`description`)
) ENGINE=MyISAM AUTO_INCREMENT=359596 DEFAULT CHARSET=utf8;

触发器：

DELIMITER ;;
CREATE TRIGGER `tblpart_insert_trigger` AFTER INSERT ON `tblpart` 
FOR EACH ROW INSERT INTO tblpart_search VALUES(NEW.partid,NEW.title,NEW.brand,NEW.description);;
DELIMITER ;

DELIMITER ;;
CREATE TRIGGER `tblpart_update_trigger` AFTER UPDATE ON `tblpart` 
FOR EACH ROW UPDATE tblpart_search SET tblpart_search.title=NEW.title,tblpart_search.brand=NEW.brand,tblpart_search.description=NEW.description WHERE tblpart_search.partid=NEW.partid;;
DELIMITER ;

DELIMITER ;;
CREATE TRIGGER `tblpart_delete_trigger` AFTER DELETE ON `tblpart` 
FOR EACH ROW DELETE FROM tblpart_search WHERE tblpart_search.partid=OLD.partid;;
DELIMITER ;

新查询：

SELECT tblpart.partid,tblpart.title,tblcategory.category,tblheadcategory.headcategory

FROM tblpart_search
INNER JOIN tblpart ON tblpart_search.partid = tblpart.partid
INNER JOIN tblcategory ON tblpart.categoryid = tblcategory.categoryid
INNER JOIN tblheadcategory ON tblcategory.headcategoryid = tblheadcategory.headcategoryid

WHERE MATCH (tblpart_search.title, tblpart_search.brand, tblpart_search.description) AGAINST ('+bmw +car' IN BOOLEAN MODE)
LIMIT 50;

将ft_min_word_len设置为3或更少，以便它可以为3 - 'BMW'和'CAR'等字符字词编制索引。

Answer 2

索引where子句中使用的字段。我不确定是否有“tblpart.title ='bmw'DESC，tblcategory.category LIKE'％bmw％'DESC”，因为我只做过“索引你的where子句中使用的字段。我不确定tblpart.title DESC，tblcategory.category DESC“

Answer 3

我认为代码 tblpart.title='bmw' DESC应该是已更改为tblpart.title LIKE '%bmw%' DESC
创建一个新表格，该表格可用作文本搜索的索引，您可以在其中存储用户输入的搜索字词以及与tblpart.title和partid相关的常用搜索字词。现在，只要用户点击搜索，您就会首先搜索此表格，以及搜索字词是否与partid的查询匹配得更快。

请帮我优化一个mysql搜索查询

更新

3 个答案: