1226关于count(*)不走主键索引反而走二级索引
Posted 秦楚风
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了1226关于count(*)不走主键索引反而走二级索引相关的知识,希望对你有一定的参考价值。
转自 http://www.2cto.com/database/201508/433975.html
- mysqlcount(*)会选哪个索引?
-
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:
123456789101112131415161718mysql>
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:
123456789101112131415161718192021222324mysql> 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:
1234567891011121314mysql>
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:
1234567891011121314151617181920212223242526mysql> 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)
1234567mysql>
select
count
(
distinct
(l_shipDATE))
from
lineitem;
+
-----------------------------+
|
count
(
distinct
(l_shipDATE)) |
+
-----------------------------+
| 2526 |
+
-----------------------------+
1 row
in
set
(0.01 sec)
1234567mysql>
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。
以上是关于1226关于count(*)不走主键索引反而走二级索引的主要内容,如果未能解决你的问题,请参考以下文章