Question

我很难在这个问题上获得更好的结果：

我有以下2个表格：

DROP TABLE IF EXISTS `casino`.`mutualfriends`;
CREATE TABLE  `casino`.`mutualfriends` (
  `CustUID` varchar(64) CHARACTER SET ascii NOT NULL,
  `CustUID2` varchar(64) CHARACTER SET ascii NOT NULL,
  `FType` tinyint(4) NOT NULL,
  PRIMARY KEY (`CustUID`,`CustUID2`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;


DROP TABLE IF EXISTS `casino`.`customers`;
CREATE TABLE  `casino`.`customers` (
  `CustFullName` varchar(45) NOT NULL,
  `CustEmail` varchar(128) NOT NULL,
  `CustUID` varchar(64) CHARACTER SET ascii NOT NULL,
  `CustMoney` bigint(20) NOT NULL DEFAULT '0',
  `SmallPicURL` varchar(120) CHARACTER SET ascii DEFAULT '',
  `LargePicURL` varchar(120) CHARACTER SET ascii DEFAULT '',
  PRIMARY KEY (`CustUID`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

customers表有1M行，共有朋友有500K行。

我很难优化这个查询;似乎有一个表扫描使用它。我想尽量减少扫描：

SELECT c.CustUID AS Cuid, c.CustFullName, c.CustMoney, c.SmallPicURL
FROM `customers` c
WHERE c.`CustUID` = '1:1542073175'
UNION
SELECT m.`CustUID2` AS Cuid, c.CustFullName, c.CustMoney, c.SmallPicURL
FROM `mutualfriends` m, `customers` c
WHERE m.`CustUID` = '1:1542073175'
AND c.`CustUID` = m.`CustUID2`
UNION
SELECT m.`CustUID` AS Cuid, c.CustFullName, c.CustMoney, c.SmallPicURL
FROM `mutualfriends` m, customers c
WHERE m.`CustUID2` = '1:1542073175'
AND c.CustUID = m.`CustUID`

Answer 1

我认为你需要MutualFriends.CustUID2的索引（重复索引，不是唯一索引;而不是主键的一部分）。主键可能为您首先列出已识别客户（'1：1542073175'）的查询提供可用索引。

检查查询计划的说明。

我认为你对UNION的第三部分进行顺序扫描是正确的，也可能是第二部分。

Answer 2

尝试此查询

SELECT c.CustUID AS Cuid, c.CustFullName, c.CustMoney, c.SmallPicURL
FROM `customers` c INNER JOIN `mutualfriends` m
IN c.CustUID = m.`CustUID`
WHERE m.`CustUID2` = '1:1542073175'

Answer 3

尝试使用此查询：

select *
from customers
where CustUID in (
  select CustUID
  from mutual_friends
  where CustUID2 = '1:1542073175'
  union all
  select CustUID2
  from mutual_friends
  where CustUID = '1:1542073175'
  union all
  select '1:1542073175'
)

那更快吗？

很难优化查询

3 个答案: