Question

我有以下经常使用的查询：

SELECT a.col1,
  b.col1,
  d.col1,
  c.col1,
  c.col2,
  c.col3,
  c.col4,
  a.col2,
  c.col5,
  c.col6,
  c.col7,
  b.col2
  FROM a
  LEFT JOIN c ON a.col3 = c.col1
  LEFT JOIN b ON a.col4 = b.col1
  LEFT JOIN d ON b.col3 = d.col2
  LEFT JOIN e ON b.col3 = e.col1
  where a.col4 != 'temp' and a.col5!=2
  GROUP BY a.col1,
  b.col1,
  d.col1,
  c.col1,
  c.col3,
  c.col4,
  a.col2,
  c.col5,
  c.col6,
  c.col7,
  b.col2
  limit 50;

此查询在远程服务器上运行，大约需要5分钟（我的网络连接速度不慢）才能显示结果。到目前为止我只使用了基本的MySQL，我不知道如何优化上述查询。我在网上搜索优化它，比如添加索引，但我发现它们仅用于非常简单的情况，我无法将它们扩展到此查询。

有人可以帮我创建索引以优化上述查询或任何其他方法以使查询运行得更快（比如创建一个额外的临时表）。

a有大约130万条记录，b - 80k条记录，c - 150k条记录，d - 150条记录。

运行查询SHOW CREATE TABLE a会得到以下结果：

| a  | CREATE TABLE `a` (
  `col1` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `col4` int(10) unsigned NOT NULL DEFAULT '0',
  `col5` int(10) unsigned NOT NULL DEFAULT '0',
  `col6` varchar(100) NOT NULL DEFAULT '',
  `col3` int(10) unsigned NOT NULL DEFAULT '0',
  `col7` varchar(250) NOT NULL DEFAULT '',
  `col2` int(10) unsigned NOT NULL DEFAULT '0',
  `col8` mediumtext,
  `col9` smallint(6) NOT NULL DEFAULT '0',
  `col10` smallint(6) NOT NULL DEFAULT '0',
  `col11` varchar(15) NOT NULL DEFAULT '',
  `col12` smallint(5) unsigned NOT NULL DEFAULT '0',
  `col13` smallint(6) NOT NULL DEFAULT '0',
  `col14` smallint(5) unsigned NOT NULL DEFAULT '0',
  `col15` smallint(5) unsigned NOT NULL DEFAULT '0',
  `col16` int(10) unsigned NOT NULL DEFAULT '0',
  `col17` int(10) unsigned NOT NULL DEFAULT '0',
  PRIMARY KEY (`col1`),
  KEY `col3` (`col3`),
  KEY `col4` (`col4`,`col3`),
  KEY `col2` (`col2`),
  KEY `col1` (`col1`),
  KEY `col1_2` (`col1`),
  KEY `col1_3` (`col1`),
  KEY `col1_4` (`col1`),
  KEY `col1_5` (`col1`),
  KEY `col1_6` (`col1`),
  KEY `col1_7` (`col1`),
  FULLTEXT KEY `col7` (`col7`,`col8`)
) ENGINE=InnoDB AUTO_INCREMENT=1339383 DEFAULT CHARSET=latin1 |

EXPLAIN <query>给出以下结果：

+----+-------------+-------+--------+---------------+---------+---------+---------------+------+----------------------------------------------+
| id | select_type | table | type   | possible_keys | key     | key_len | ref           | rows | Extra                                        |
+----+-------------+-------+--------+---------------+---------+---------+---------------+------+----------------------------------------------+
|  1 | SIMPLE      | e     | ALL    | NULL          | NULL    | NULL    | NULL          |  149 | Using where; Using temporary; Using filesort |
|  1 | SIMPLE      | b     | ref    | PRIMARY,col3  | col3    | 2       | db.e.col1     |  286 | Using index condition                        |
|  1 | SIMPLE      | d     | eq_ref | PRIMARY       | PRIMARY | 2       | db.b.col3     |    1 | NULL                                         |
|  1 | SIMPLE      | a     | ref    | col4          | col4    | 4       | db.b.col1     |   10 | Using where                                  |
|  1 | SIMPLE      | c     | eq_ref | PRIMARY       | PRIMARY | 4       | db.a.col3     |    1 | NULL                                         |
+----+-------------+-------+--------+---------------+---------+---------+---------------+------+----------------------------------------------+

Answer 1

最低指数::

library(devtools)
 install_github("repo/SparkR-pkg", ref="branchname", subdir="pkg")

您也可以创建，

    CREATE INDEX a0 ON a (col4, col5) ;
    CREATE INDEX a1 ON a (col3) ;
    CREATE INDEX a2 ON a (col4) ;

    CREATE INDEX b1 ON b (col1) ;
    CREATE INDEX b2 ON b (col3) ;

    CREATE INDEX c1 ON c (col1) ;

    CREATE INDEX d1 ON d (col2) ;

使用许多LEFT JOIN和GROUP BY优化MySQL查询

1 个答案: