连接索引字段不使用索引

Question

我有两张桌子.我写了一个查询来在一列上加入它们.此列在两个表中都已编制索引,但MySQL未使用索引.有人a)告诉我为什么和b)告诉我如何让MySQL使用索引快速加入这些表.

第一张表:

CREATE TABLE `dol_msa_zip_assoc` (
  `pkey` int(5) unsigned NOT NULL AUTO_INCREMENT,
  `zip` varchar(5) NOT NULL DEFAULT '',
  `pmsa_msa` varchar(5) NOT NULL DEFAULT '',
  PRIMARY KEY (`pkey`),
  KEY `zip` (`zip`)
) ENGINE=InnoDB  DEFAULT CHARSET=utf8;

它包含42690条记录.

另一张桌子:

CREATE TABLE `v3_msa_zip_assoc` (
  `pkey` int(5) unsigned NOT NULL AUTO_INCREMENT,
  `zip` varchar(9) NOT NULL DEFAULT '',
  `pmsa_msa` varchar(6) NOT NULL DEFAULT '',
  PRIMARY KEY (`pkey`),
  KEY `zip` (`zip`)
) ENGINE=MyISAM  DEFAULT CHARSET=latin1;

它包含42486条记录.

我的查询,旨在查找第一个表中不在第二个表中的记录:

SELECT d.*, o.* FROM `dol_msa_zip_assoc` d
LEFT JOIN `v3_msa_zip_assoc` o
ON o.zip = d.zip
WHERE o.zip IS NULL

当我解释此查询时,我看到zip列上的索引未被使用:

+----+-------------+-------+------+---------------+------+---------+------+-------+----------------------------------------------------------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows  | Extra                                                          |
+----+-------------+-------+------+---------------+------+---------+------+-------+----------------------------------------------------------------+
|  1 | SIMPLE      | d     | ALL  | NULL          | NULL | NULL    | NULL | 42915 | NULL                                                           |
|  1 | SIMPLE      | o     | ALL  | NULL          | NULL | NULL    | NULL | 42486 | Using where; Not exists; Using join buffer (Block Nested Loop) |
+----+-------------+-------+------+---------------+------+---------+------+-------+----------------------------------------------------------------+

导致此失败的原因可能是什么？是因为一个zip列是latin1而另一个是utf8？如何使用此查询来使用索引,以便它不需要运行年龄？

编辑:反转JOIN的顺序显然确实利用了索引:

SELECT d . * , o . *
FROM `v3_msa_zip_assoc` o
LEFT JOIN `dol_msa_zip_assoc` d ON d.zip = o.zip
WHERE d.zip IS NULL

这是解释:

+----+-------------+-------+------+---------------+------+---------+------+-------+-------------------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows  | Extra                   |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------------------+
|  1 | SIMPLE      | o     | ALL  | NULL          | NULL | NULL    | NULL | 42486 | NULL                    |
|  1 | SIMPLE      | d     | ref  | zip           | zip  | 17      | func |     1 | Using where; Not exists |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------------------+

编辑2:我已经改变旧表的结构以使用utf8排序规则.现在定义如下:

CREATE TABLE `v3_msa_zip_assoc` (
  `pkey` int(5) unsigned NOT NULL AUTO_INCREMENT,
  `zip` varchar(9) NOT NULL DEFAULT '',
  `pmsa_msa` varchar(6) NOT NULL DEFAULT '',
  PRIMARY KEY (`pkey`),
  KEY `zip` (`zip`)
) ENGINE=InnoDB  DEFAULT CHARSET=utf8;

这似乎解决了这个问题:

mysql> EXPLAIN SELECT d . * , o . *
    -> FROM `dol_msa_zip_assoc` d
    -> LEFT JOIN `v3_msa_zip_assoc` o ON o.zip = d.zip
    -> WHERE o.zip IS NULL;
+----+-------------+-------+------+---------------+------+---------+---------------------+-------+-------------------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref                 | rows  | Extra                   |
+----+-------------+-------+------+---------------+------+---------+---------------------+-------+-------------------------+
|  1 | SIMPLE      | d     | ALL  | NULL          | NULL | NULL    | NULL                | 42915 | NULL                    |
|  1 | SIMPLE      | o     | ref  | zip           | zip  | 29      | myplan_v4_dev.d.zip |     1 | Using where; Not exists |
+----+-------------+-------+------+---------------+------+---------+---------------------+-------+-------------------------+

Answer 1

这是阻止使用索引的字符集.

其中一个zip列是latin1,另一个是utf8.

有一个隐含的转换,相当于CONVERT(zip USING charset)一方面,