MySQL没有使用ORDER BY的索引

Question

我有一个名为'test'的简单MySQL表,有两列:

1. 自动递增int列名为'id'
2. Varchar(3000)专栏名为'textcol'

我根据'textcol'列在表中创建了一个索引.但是,ORDER BY查询似乎没有使用索引,即在textcol上使用ORDER BY的简单查询上的EXPLAIN语句在其输出的Key列中显示NULL,并且还使用filesort.

任何用于进行更改以帮助使用ORDER by query的索引的指针对我都很有用.

MySQL版本由"mysql --version"命令给出:

mysql Ver 14.14 Distrib 5.1.58,debian-linux-gnu(x86_64)使用readline 6.2

mysql> CREATE TABLE test (id INTEGER NOT NULL AUTO_INCREMENT, PRIMARY KEY(id), textcol VARCHAR(3000));
Query OK, 0 rows affected (0.05 sec)

mysql> DESCRIBE test;
+---------+---------------+------+-----+---------+----------------+
| Field   | Type          | Null | Key | Default | Extra          |
+---------+---------------+------+-----+---------+----------------+
| id      | int(11)       | NO   | PRI | NULL    | auto_increment |
| textcol | varchar(3000) | YES  |     | NULL    |                |
+---------+---------------+------+-----+---------+----------------+
2 rows in set (0.00 sec)

mysql> CREATE INDEX textcolindex ON test (textcol);
Query OK, 0 rows affected, 2 warnings (0.06 sec)
Records: 0  Duplicates: 0  Warnings: 0

mysql> SHOW INDEX FROM test;
+-------+------------+--------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+
| Table | Non_unique | Key_name     | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment |
+-------+------------+--------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+
| test  |          0 | PRIMARY      |            1 | id          | A         |           0 |     NULL | NULL   |      | BTREE      |         |
| test  |          1 | textcolindex |            1 | textcol     | A         |        NULL |     1000 | NULL   | YES  | BTREE      |         |
+-------+------------+--------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+
2 rows in set (0.00 sec)

mysql> INSERT INTO test (textcol) VALUES ('test1');
Query OK, 1 row affected (0.00 sec)

mysql> INSERT INTO test (textcol) VALUES ('test2');
Query OK, 1 row affected (0.00 sec)

mysql> INSERT INTO test (textcol) VALUES ('test3');
Query OK, 1 row affected (0.00 sec)

mysql> INSERT INTO test (textcol) VALUES ('test4');
Query OK, 1 row affected (0.00 sec)


mysql> EXPLAIN SELECT * FROM test ORDER BY textcol;
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows | Extra          |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
|  1 | SIMPLE      | test  | ALL  | NULL          | NULL | NULL    | NULL |    4 | Using filesort |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
1 row in set (0.00 sec)

mysql> EXPLAIN SELECT * FROM test ORDER BY id;
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows | Extra          |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
|  1 | SIMPLE      | test  | ALL  | NULL          | NULL | NULL    | NULL |    4 | Using filesort |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
1 row in set (0.00 sec)

Answer 1

由于它必须加载整个表来回答查询并且排序4个元素很便宜,查询优化器可能只是避免触及索引.大表还会发生吗？

请注意,varchar(3000)列不能是覆盖索引,因为MySQL不会包含多于索引中varchar的前768个字节.

如果您希望查询仅读取索引,则索引必须包含您要SELECT在其中的每个列.在innodb上,一旦你使textcol足够小,它应该开始为你的双列表工作; 在MyISAM上,您需要自己包含主键列,例如CREATE INDEX textcolindex ON test (textcol,id);

Answer 2

我遇到了同样的问题。MySQL是愚蠢的。仅供参考：我有一张包含超过 500,000,000 条记录的表。我想：

select * from table order by tid limit 10000000, 10;

tid 是表中的主键，由 mysql 自动索引。

这花了很长时间，我取消了查询。然后我让mysql“解释”查询并认识到它不会使用主键的索引。从 mysql 阅读了许多文档后，我试图通过“USE INDEX(...)”强制 mysql 使用索引，但 dis 也不起作用。然后我意识到mysql似乎总是将where子句与order by子句相关联。因此，我尝试使用触及索引的条件扩展 where 子句。我结束了：

select * from table use index (PRIMARY) where tid > 0 order by tid limit 10000000, 10;

其中 tid 是表中的主键，是从 1 开始的自动增量值。

在我让 mysql 向我解释查询之后，这奏效了。看哪：查询只用了 4 秒。

Answer 3

有关ORDER BY优化的一些有用的文章:

http://www.mysqlperformanceblog.com/2006/09/01/order-by-limit-performance-optimization/

http://opsmonkey.blogspot.co.uk/2009/03/mysql-query-optimization-for-order-by.html

如上所述,将varchar保持在767并为订单添加一个键:

CREATE TABLE test (
id INTEGER NOT NULL AUTO_INCREMENT,
textcol VARCHAR(767),
PRIMARY KEY(id),
KEY orderby (`textcol`)
);

要避免filesorts添加额外的'WHERE'参数,请使用多列索引扩展'orderby'索引键:

CREATE TABLE test (
id INTEGER NOT NULL AUTO_INCREMENT, 
tom INT(11) NOT NULL DEFAULT 0,
gerry INT(11) NOT NULL DEFAULT 0,
textcol VARCHAR(767),
PRIMARY KEY(id), 
KEY orderby (`tom`,`gerry`, `textcol`)
);

也:

INSERT INTO test (tom, gerry, textcol) VALUES (1,2,'test4');
INSERT INTO test (tom, gerry, textcol) VALUES (1,2,'test2');
EXPLAIN SELECT id, textcol FROM test WHERE tom = 1 AND gerry =2 ORDER BY textcol;

额外:'使用地点; 使用索引'