1226关于count(*)不走主键索引反而走二级索引

转自 http://www.2cto.com/database/201508/433975.html

mysqlcount(*)会选哪个索引？

2015-08-19      0个评论    来源：Database、Code  

收藏    我要投稿

 

今天在查询一个表行数的时候，发现count(1)和count(*)执行效率居然是一样的。这跟Oracle还是有区别的。遂查看两种方式的执行计划：

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

mysql> select count(1) from customer; +----------+ | count(1) | +----------+ |   150000 | +----------+ 1 row in set (0.03 sec)   mysql> flush tables; Query OK, 0 rows affected (0.00 sec)   mysql> select count(*) from customer; +----------+ | count(*) | +----------+ |   150000 | +----------+ 1 row in set (0.03 sec)

查看执行计划：

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

mysql> explain select count(1) from customer; +----+-------------+----------+-------+---------------+---------------+---------+------+--------+-------------+ | id | select_type | table    | type  | possible_keys | key           | key_len | ref  | rows   | Extra       | +----+-------------+----------+-------+---------------+---------------+---------+------+--------+-------------+ |  1 | SIMPLE      | customer | index | NULL          | i_c_nationkey | 5       | NULL | 151191 | Using index | +----+-------------+----------+-------+---------------+---------------+---------+------+--------+-------------+ 1 row in set (0.00 sec)   mysql> explain select count(*) from customer; +----+-------------+----------+-------+---------------+---------------+---------+------+--------+-------------+ | id | select_type | table    | type  | possible_keys | key           | key_len | ref  | rows   | Extra       | +----+-------------+----------+-------+---------------+---------------+---------+------+--------+-------------+ |  1 | SIMPLE      | customer | index | NULL          | i_c_nationkey | 5       | NULL | 151191 | Using index | +----+-------------+----------+-------+---------------+---------------+---------+------+--------+-------------+ 1 row in set (0.00 sec)   mysql> show index from customer; +----------+------------+---------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+ | Table    | Non_unique | Key_name      | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment | +----------+------------+---------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+ | customer |          0 | PRIMARY       |            1 | c_custkey   | A         |      150525 |     NULL | NULL   |      | BTREE      |         |               | | customer |          1 | i_c_nationkey |            1 | c_nationkey | A         |          47 |     NULL | NULL   | YES  | BTREE      |         |               | +----------+------------+---------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+ 2 rows in set (0.08 sec)

发现不管是count(1)或count(*)都是走的i_c_nationkey这个索引。平时我们检索数据的时候肯定是主键索引效率高，那么我们强制主键索引来看看：

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14

mysql> select count(*) from customer force index(PRIMARY); +----------+ | count(*) | +----------+ |   150000 | +----------+ 1 row in set (0.68 sec) mysql> explain select count(*) from customer force index(PRIMARY); +----+-------------+----------+-------+---------------+---------+---------+------+--------+-------------+ | id | select_type | table    | type  | possible_keys | key     | key_len | ref  | rows   | Extra       | +----+-------------+----------+-------+---------------+---------+---------+------+--------+-------------+ |  1 | SIMPLE      | customer | index | NULL          | PRIMARY | 4       | NULL | 150525 | Using index | +----+-------------+----------+-------+---------------+---------+---------+------+--------+-------------+ 1 row in set (0.00 sec)

可以看到走主键索引的时候效率比较差。那么是为什么呢。
平时我们检索一列的时候，基本上等值或范围查询，那么索引基数大的索引必然效率很高。但是在做count(*)的时候并没有检索具体的一行或者一个范围。那么选择基数小的索引对
count操作效率会更高。在做count操作的时候，mysql会遍历每个叶子节点，所以基数越小，效率越高。mysql非聚簇索引叶子节点保存的主键ID，所以需要检索两遍索引。但是这里相对于遍历主键索引。及时检索两遍索引效率也比单纯的检索主键索引快。
那么再以一个表作为证明：

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

mysql> explain select count(*) from lineitem; +----+-------------+----------+-------+---------------+--------------+---------+------+---------+-------------+ | id | select_type | table    | type  | possible_keys | key          | key_len | ref  | rows    | Extra       | +----+-------------+----------+-------+---------------+--------------+---------+------+---------+-------------+ |  1 | SIMPLE      | lineitem | index | NULL          | i_l_shipdate | 4       | NULL | 6008735 | Using index | +----+-------------+----------+-------+---------------+--------------+---------+------+---------+-------------+ 1 row in set (0.00 sec)   mysql> show index from lineitem; +----------+------------+-----------------------+--------------+---------------+-----------+-------------+----------+--------+------+------------+---------+---------------+ | Table    | Non_unique | Key_name              | Seq_in_index | Column_name   | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment | +----------+------------+-----------------------+--------------+---------------+-----------+-------------+----------+--------+------+------------+---------+---------------+ | lineitem |          0 | PRIMARY               |            1 | l_orderkey    | A         |     2997339 |     NULL | NULL   |      | BTREE      |         |               | | lineitem |          0 | PRIMARY               |            2 | l_linenumber  | A         |     5994679 |     NULL | NULL   |      | BTREE      |         |               | | lineitem |          1 | i_l_shipdate          |            1 | l_shipDATE    | A         |        5208 |     NULL | NULL   | YES  | BTREE      |         |               | | lineitem |          1 | i_l_suppkey_partkey   |            1 | l_partkey     | A         |      428191 |     NULL | NULL   | YES  | BTREE      |         |               | | lineitem |          1 | i_l_suppkey_partkey   |            2 | l_suppkey     | A         |     1998226 |     NULL | NULL   | YES  | BTREE      |         |               | | lineitem |          1 | i_l_partkey           |            1 | l_partkey     | A         |      461129 |     NULL | NULL   | YES  | BTREE      |         |               | | lineitem |          1 | i_l_suppkey           |            1 | l_suppkey     | A         |       19213 |     NULL | NULL   | YES  | BTREE      |         |               | | lineitem |          1 | i_l_receiptdate       |            1 | l_receiptDATE | A         |          17 |     NULL | NULL   | YES  | BTREE      |         |               | | lineitem |          1 | i_l_orderkey          |            1 | l_orderkey    | A         |     2997339 |     NULL | NULL   |      | BTREE      |         |               | | lineitem |          1 | i_l_orderkey_quantity |            1 | l_orderkey    | A         |     1998226 |     NULL | NULL   |      | BTREE      |         |               | | lineitem |          1 | i_l_orderkey_quantity |            2 | l_quantity    | A         |     5994679 |     NULL | NULL   | YES  | BTREE      |         |               | | lineitem |          1 | i_l_commitdate        |            1 | l_commitDATE  | A         |        7836 |     NULL | NULL   | YES  | BTREE      |         |               | +----------+------------+-----------------------+--------------+---------------+-----------+-------------+----------+--------+------+------------+---------+---------------+ 12 rows in set (0.96 sec)

这里一看l_shipDATE并不是基数最小的呀，殊不知这个统计信息是不准确的。我们用sql看一下。

?

1 2 3 4 5 6 7

mysql> select count(distinct(l_shipDATE)) from lineitem; +-----------------------------+ | count(distinct(l_shipDATE)) | +-----------------------------+ |                        2526 | +-----------------------------+ 1 row in set (0.01 sec)

那么比他小的那些列呢？

?

1 2 3 4 5 6 7

mysql> select count(distinct(l_receiptDATE)) from lineitem; +--------------------------------+ | count(distinct(l_receiptDATE)) | +--------------------------------+ |                           2554 | +--------------------------------+ 1 row in set (0.01 sec)

其他就不看了，这里再次说明mysql选择了基数小的索引。