MySQL key_len 大小的计算

时间:2023-03-09 20:30:27
MySQL key_len 大小的计算

背景:

当用Explain查看SQL的执行计划时,里面有列显示了 key_len 的值,根据这个值可以判断索引的长度,在组合索引里面可以更清楚的了解到了哪部分字段使用到了索引。

环境:

CREATE TABLE `tmp_0612` (
`id` int(11) NOT NULL,
`name` varchar(10) DEFAULT NULL,
`age` int(11) DEFAULT NULL,
`address` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8

插入数据:

insert into tmp_0612 values(1,'a',11,'hz'),(2,'b',22,'gz'),(3,'c',33,'aa');

创建索引:

alter table tmp_0612 add index idx_name(name);
alter table tmp_0612 add index idx_age(age);

测试:

explain select * from tmp_0612 where name ='a';
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| 1 | SIMPLE | tmp_0612 | ref | idx_name | idx_name | 33 | const | 1 | Using index condition |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+

从上面的执行计划可知,索引的长度是33。比预想的30(10*3(utf8))要高出3字节,为什么呢?进一步测试:

修改name 成 not null

alter table tmp_0612 modify name varchar(10) not null;

再看执行计划:

explain select * from tmp_0612 where name ='a';
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| 1 | SIMPLE | tmp_0612 | ref | idx_name | idx_name | 32 | const | 1 | Using index condition |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+

发现上面的执行计划和第一次的有区别(key_len),经过多次测试,发现字段允许NULL的会多出一个字节。想到了之前的一篇文章,NULL是需要一个标志位的,占用1个字符。那还有2个字节怎么算?这里想到的是会不会和 多字节字符集 相关?那改字符集试试:

alter table tmp_0612 convert to charset latin1;

explain select * from tmp_0612 where name ='a';
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| 1 | SIMPLE | tmp_0612 | ref | idx_name | idx_name | 12 | const | 1 | Using index condition |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+

发现还是多了2个字节,看来和多字节字符集没有关系了。那会不会和 变长字段 有关系?再试试:

alter table tmp_0612 convert to charset utf8;

alter table tmp_0612 modify name char(10) not null;

explain select * from tmp_0612 where name ='a';
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| 1 | SIMPLE | tmp_0612 | ref | idx_name | idx_name | 30 | const | 1 | Using index condition |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+

和预料的一样了,是30=10*3。到这里相信大家都已经很清楚了,要是还比较模糊就看反推到33字节。

改成允许NULL,应该会变成31。

alter table tmp_0612 modify name char(10);

explain select * from tmp_0612 where name ='a';
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| 1 | SIMPLE | tmp_0612 | ref | idx_name | idx_name | 31 | const | 1 | Using index condition |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+

改成变长字段类型,应该会变成33。

alter table tmp_0612 modify name varchar(10);

explain select * from tmp_0612 where name ='a';
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| 1 | SIMPLE | tmp_0612 | ref | idx_name | idx_name | 33 | const | 1 | Using index condition |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+

改成单字节字符集,还是还是需要额外的3字节(1:null ;变长字段:2),和字符集无关。

alter table tmp_0612 convert to charset latin1;

explain select * from tmp_0612 where name ='a';
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+
| 1 | SIMPLE | tmp_0612 | ref | idx_name | idx_name | 13 | const | 1 | Using index condition |
+----+-------------+----------+------+---------------+----------+---------+-------+------+-----------------------+

反推上去都和预测的一样。

其他测试:

explain select * from tmp_0612 where age = 11;
+----+-------------+----------+------+---------------+---------+---------+-------+------+-------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+---------+---------+-------+------+-------+
| 1 | SIMPLE | tmp_0612 | ref | idx_age | idx_age | 5 | const | 1 | NULL |
+----+-------------+----------+------+---------------+---------+---------+-------+------+-------+ alter table tmp_0612 modify age int not null; explain select * from tmp_0612 where age = 11;
+----+-------------+----------+------+---------------+---------+---------+-------+------+-------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+----------+------+---------------+---------+---------+-------+------+-------+
| 1 | SIMPLE | tmp_0612 | ref | idx_age | idx_age | 4 | const | 1 | NULL |
+----+-------------+----------+------+---------------+---------+---------+-------+------+-------+

int 是占4个字符的,上面key_len的也是符合预期。关于组合索引的,可以自己去测试玩。

总结:

变长字段需要额外的2个字节,固定长度字段不需要额外的字节。而null都需要1个字节的额外空间,所以以前有个说法:索引字段最好不要为NULL,因为NULL让统计更加复杂,并且需要额外一个字节的存储空间。

key_len的长度计算公式:

varchr(10)变长字段且允许NULL      : 10*(Character Set:utf8=3,gbk=2,latin1=1)+1(NULL)+2(变长字段)
varchr(10)变长字段且不允许NULL : 10*(Character Set:utf8=3,gbk=2,latin1=1)+2(变长字段) char(10)固定字段且允许NULL : 10*(Character Set:utf8=3,gbk=2,latin1=1)+1(NULL)
char(10)固定字段且不允许NULL : 10*(Character Set:utf8=3,gbk=2,latin1=1)