MySQL查询性能帮助，许多相同的表被连接

Question

我正在编写一个创建SQL查询的PHP脚本。该脚本和数据库用于Joomla CMS，特别是查询SOBIPro组件的表(以使用在该组件中输入的数据)。但是，由于SOBI Pro表的处理方式，字段的每个实例都是表中自己的行，这意味着对于我要拉回的每个字段，都要包含一个单独的表实例。这似乎不是很有效，实际上在这个搜索中它超时了。

SQL查询如下所示(这是由我的PHP代码生成的)：

    SELECT DISTINCT o.id AS entryid, o.parent AS parentID, name.baseData AS title,business.baseData AS business_data,
    contact_fn.baseData AS contact_fn_data ,contact_ln.baseData AS contact_ln_data ,position.baseData AS position_data,
    civic1.baseData AS civic1_data ,civic2.baseData AS civic2_data ,mailing.baseData AS mailing_data,
    community.baseData AS community_data ,municip.baseData AS municip_data ,county.baseData AS county_data,
    province.baseData AS province_data ,country.baseData AS country_data ,postal.baseData AS descr_data,
    phone.baseData AS phone_data ,tollfree.baseData AS tollfree_data ,fax.baseData AS fax_data,
    email.baseData AS email_data ,web.baseData AS web_data ,empTotal.baseData AS empTotal_data
    FROM jos_sobipro_object AS o
    INNER JOIN jos_sobipro_field_data AS name ON name.sid = o.id
    INNER JOIN jos_sobipro_relations AS r ON o.id = r.id
    LEFT JOIN jos_sobipro_field_data AS business ON business.sid = o.id AND business.fid = 36
    LEFT JOIN jos_sobipro_field_data AS contact_fn ON contact_fn.sid = o.id AND contact_fn.fid = 74
    LEFT JOIN jos_sobipro_field_data AS contact_ln ON contact_ln.sid = o.id AND contact_ln.fid = 75
    LEFT JOIN jos_sobipro_field_data AS position ON position.sid = o.id AND position.fid = 76
    LEFT JOIN jos_sobipro_field_data AS civic1 ON civic1.sid = o.id AND civic1.fid = 77
    LEFT JOIN jos_sobipro_field_data AS civic2 ON civic2.sid = o.id AND civic2.fid = 78
    LEFT JOIN jos_sobipro_field_data AS mailing ON mailing.sid = o.id AND mailing.fid = 79
    LEFT JOIN jos_sobipro_field_data AS community ON community.sid = o.id AND community.fid = 80
    LEFT JOIN jos_sobipro_field_data AS municip ON municip.sid = o.id AND municip.fid = 81
    LEFT JOIN jos_sobipro_field_data AS county ON county.sid = o.id AND county.fid = 82
    LEFT JOIN jos_sobipro_field_data AS province ON province.sid = o.id AND province.fid = 83
    LEFT JOIN jos_sobipro_field_data AS country ON country.sid = o.id AND country.fid = 84
    LEFT JOIN jos_sobipro_field_data AS postal ON postal.sid = o.id AND postal.fid = 85
    LEFT JOIN jos_sobipro_field_data AS phone ON phone.sid = o.id AND phone.fid = 86
    LEFT JOIN jos_sobipro_field_data AS tollfree ON tollfree.sid = o.id AND tollfree.fid = 87
    LEFT JOIN jos_sobipro_field_data AS fax ON fax.sid = o.id AND fax.fid = 88
    LEFT JOIN jos_sobipro_field_data AS email ON email.sid = o.id AND email.fid = 89
    LEFT JOIN jos_sobipro_field_data AS web ON web.sid = o.id AND web.fid = 90
    LEFT JOIN jos_sobipro_field_data AS empTotal ON empTotal.sid = o.id AND empTotal.fid = 106
    WHERE o.approved = 1 AND o.oType = 'entry' AND name.fid = 36 AND name.baseData <> ''
    AND name.section = 54 AND r.pid IN (415,418,425,431,458) AND (municip.baseData = "Municipality Name")
    ORDER BY name.baseData ASC

只要不涉及municip.baseData搜索，它似乎工作得很快，在这种情况下，即使目录中有15个条目，它也会失败。必须有一种更好的方法来设计这段SQL代码，同时仍然返回所需的所有字段。这个查询是通过AJAX调用的，最终目录中会有2000+条目。

编辑:根据请求，以下是解释输出：

    id  select_type     table   type    possible_keys   key     key_len     ref     rows    Extra
    1   SIMPLE  name    ref     PRIMARY     PRIMARY     8   const,const     15  Using where; Using temporary; Using filesort
    1   SIMPLE  municip     ref     PRIMARY     PRIMARY     4   const   9   Using where
    1   SIMPLE  o   eq_ref  PRIMARY,oType   PRIMARY     4   [[dbname]].municip.sid  1   Using where
    1   SIMPLE  county  ref     PRIMARY     PRIMARY     4   const   10  
    1   SIMPLE  province    ref     PRIMARY     PRIMARY     4   const   10  
    1   SIMPLE  country     ref     PRIMARY     PRIMARY     4   const   8   
    1   SIMPLE  postal  ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  business    ref     PRIMARY     PRIMARY     4   const   15  
    1   SIMPLE  contact_fn  ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  contact_ln  ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  position    ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  civic1  ref     PRIMARY     PRIMARY     4   const   10  
    1   SIMPLE  civic2  ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  phone   ref     PRIMARY     PRIMARY     4   const   10  
    1   SIMPLE  tollfree    ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  fax     ref     PRIMARY     PRIMARY     4   const   10  
    1   SIMPLE  email   ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  mailing     ref     PRIMARY     PRIMARY     4   const   11  
    1   SIMPLE  community   ref     PRIMARY     PRIMARY     4   const   9   
    1   SIMPLE  web     ref     PRIMARY     PRIMARY     4   const   10  
    1   SIMPLE  empTotal    ref     PRIMARY     PRIMARY     4   const   10  
    1   SIMPLE  r   ref     PRIMARY     PRIMARY     4   [[dbname]].name.sid     3   Using where; Using index; Distinct

Answer 1

很多时候，当您有一个过度扩展的JOIN/etc时，SQL引擎将会挂起，试图找到较小的结果集，并以一种效率较低的方式回填链接。您的查询看起来不错。

您的主表(来自jos_sobipro_object AS o)实际上是查询的关键驱动元素。尝试使用MySQL添加"STRAIGHT_JOIN“特殊关键字..

选择STRAIGHT_JOIN DISTINCT ...查询的其余部分...

STRAIGHT_JOIN告诉优化器按照我在这里列出的顺序执行查询。如果知道第一个表是用于查询数据的主要表，那么它的运行速度会更快。

也就是说，并不是确切地看到索引信息，我会特别在jos_sobipro_field_data上有一个索引，通过(SID，FID)获得“查找”数据。

我不得不用类似的方法处理主表中14+百万条记录的政府数据，并连接到22+查找表。MySQL会在30+小时后挂起。通过添加STRAIGHT_JOIN，查询在大约3小时内完成(正如它所做的那样)。