有时我们想统计某种类型有多少个，会用这个SQL。全表扫描之余，还要filesort，耗时1.34秒。

mysql>  select country,count(*) from t1 group by country;
+---------+----------+
| country | count(*) |
+---------+----------+
| NULL    |       32 |
| africa  |   524288 |
| america |   524288 |
| china   |   524288 |
+---------+----------+
4 rows in set (1.34 sec)

mysql> desc select country,count(*) from t1 group by country;
+----+-------------+-------+------+---------------+------+---------+------+---------+---------------------------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows    | Extra                           |
+----+-------------+-------+------+---------------+------+---------+------+---------+---------------------------------+
|  1 | SIMPLE      | t1    | ALL  | NULL          | NULL | NULL    | NULL | 1573382 | Using temporary; Using filesort |
+----+-------------+-------+------+---------------+------+---------+------+---------+---------------------------------+
1 row in set (0.00 sec)

以下是两种优化方法，都是全表扫描，但估计count()要比sum()耗的CPU少点，方法更佳。

mysql> select count(country='africa' or null) as africa,count(country='america' or null) as america, count(country='china' or null) as china from t1;
+--------+---------+--------+
| africa | america | china  |
+--------+---------+--------+
| 524288 |  524288 | 524288 |
+--------+---------+--------+
1 row in set (0.78 sec)

mysql> desc select count(country='africa' or null) as africa,count(country='america' or null) as america, count(country='china' or null) as china from t1;
+----+-------------+-------+------+---------------+------+---------+------+---------+-------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows    | Extra |
+----+-------------+-------+------+---------------+------+---------+------+---------+-------+
|  1 | SIMPLE      | t1    | ALL  | NULL          | NULL | NULL    | NULL | 1573382 |       |
+----+-------------+-------+------+---------------+------+---------+------+---------+-------+
1 row in set (0.00 sec)



mysql> select sum(country='africa') as africa ,sum(country='america') as america,sum(country='china') from t1;
+--------+---------+----------------------+
| africa | america | sum(country='china') |
+--------+---------+----------------------+
| 524288 |  524288 |               524288 |
+--------+---------+----------------------+
1 row in set (0.86 sec)

mysql> desc select sum(country='africa') as africa ,sum(country='america') as america,sum(country='china') from t1;
+----+-------------+-------+------+---------------+------+---------+------+---------+-------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows    | Extra |
+----+-------------+-------+------+---------------+------+---------+------+---------+-------+
|  1 | SIMPLE      | t1    | ALL  | NULL          | NULL | NULL    | NULL | 1573382 |       |
+----+-------------+-------+------+---------------+------+---------+------+---------+-------+
1 row in set (0.00 sec)

版本mysql5.5.30

Mysql——SQL优化-统计某种类型的个数