天天看点

MySQL执行计划explain的key_len解析

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。下面演示中,表结构的合理性这边暂且不说,只是证明一下索引长度的计算方法。目前大部分博文是字符类型的索引长度计算方法,下面列举几个类型的索引长度计算方法:

1、整数类型

(dg1)[email protected] [mytest]> desc table_key;

+---------+-------------+------+-----+---------+-------+

| field   | type        | null | key | default | extra |

| id      | int(11)     | no   |     | null    |       |

| sid     | bigint(20)  | no   |     | 0       |       |

| name    | char(10)    | yes  |     | null    |       |

| age     | tinyint(4)  | yes  |     | null    |       |

| sex     | tinyint(4)  | no   |     | null    |       |

| address | varchar(10) | yes  | mul | null    |       |

rows in set (0.01 sec)

(dg1)[email protected] [mytest]>create index age_index on table_key (age);

来看看tinyint类型的索引长度,在not null 和 null 的时候 分别是1和2,tinyint字段长度为1,因为null 需要额外一个字节标记为空

(dg1)[email protected] [mytest]> explain extended select * from table_key where age=38;

+----+-------------+-----------+------+---------------+-----------+---------+-------+------+----------+-------+

| id | select_type | table     | type | possible_keys | key       | key_len | ref   | rows | filtered | extra |

|  1 | simple      | table_key | ref  | age_index     | age_index | 1       | const |    1 |   100.00 | null  |

row in set, 1 warning (0.00 sec)

(dg1)[email protected] [mytest]> alter table table_key modify age  tinyint(4);

|  1 | simple      | table_key | ref  | age_index     | age_index | 2       | const |    1 |   100.00 | null  |

(dg1)[email protected] [mytest]> 

看看bigint类型的索引长度,同样是 not null 和 null值的时候,分别是8和9,聪明的你应该知道了,bigint长度为8。

(dg1)[email protected] [mytest]> alter table table_key add key sid_index (sid);

query ok, 0 rows affected (0.04 sec)

records: 0  duplicates: 0  warnings: 0

(dg1)[email protected] [mytest]> desc select * from table_key where sid=6;

+----+-------------+-----------+------+---------------+-----------+---------+-------+------+-------+

| id | select_type | table     | type | possible_keys | key       | key_len | ref   | rows | extra |

|  1 | simple      | table_key | ref  | sid_index     | sid_index | 8       | const |    1 | null  |

row in set (0.00 sec)

(dg1)[email protected] [mytest]> alter table table_key modify sid bigint(20);

query ok, 0 rows affected (0.08 sec)

|  1 | simple      | table_key | ref  | sid_index     | sid_index | 9       | const |    1 | null  |

看看smallint类型索引长度,同样是 not null 和 null值的时候,分别是2和3 smallint长度为2,允许为空需要一个字节标记

(dg1)[email protected] [mytest]> alter table table_key modify sid smallint not null default 0;

query ok, 9 rows affected (0.04 sec)

records: 9  duplicates: 0  warnings: 0

|  1 | simple      | table_key | ref  | sid_index     | sid_index | 2       | const |    1 | null  |

(dg1)[email protected] [mytest]> alter table table_key modify sid smallint ;

query ok, 0 rows affected (0.07 sec)

|  1 | simple      | table_key | ref  | sid_index     | sid_index | 3       | const |    1 | null  |

看看mediumint类型索引长度,同样是 not null 和 null值的时候,分别是3和4

(dg1)[email protected] [mytest]> alter table table_key modify sid mediumint not null;

query ok, 0 rows affected (0.06 sec)

(dg1)[email protected] [mytest]> alter table table_key modify sid mediumint ; 

query ok, 0 rows affected (0.06 sec) 

+----+-------------+-----------+------+---------------+-----------+---------+-------+------+-------+ 

| id | select_type | table     | type | possible_keys | key       | key_len | ref   | rows | extra | 

|  1 | simple      | table_key | ref  | sid_index     | sid_index | 4       | const |    1 | null  | 

整数类型索引长度跟字段长度有关,如果允许为空,需要额外一个字节去标记为空

2.浮点数类型

表结构

create table `table_key1` (

`id`  int not null auto_increment ,

`c1`  float not null ,

`c2`  double not null ,

`c3`  decimal not null ,

`c4`  date not null ,

`c5`  timestamp not null on update current_timestamp ,

`c6`  datetime not null ,

primary key (`id`)

)

看看float类型的索引长度,not null和null的时候,分别是4和5

(dg1)[email protected] [mytest]> desc select * from table_key1 where c1 = '3.22';

+----+-------------+------------+------+---------------+----------+---------+-------+------+-----------------------+

| id | select_type | table      | type | possible_keys | key      | key_len | ref   | rows | extra                 |

|  1 | simple      | table_key1 | ref  | c1_index      | c1_index | 4       | const |    8 | using index condition |

(dg1)[email protected] [mytest]> alter table table_key1 modify c1 float;

query ok, 0 rows affected (0.05 sec)

|  1 | simple      | table_key1 | ref  | c1_index      | c1_index | 5       | const |    8 | using index condition |

看看double类型的索引长度,not null和null的时候,分别是8和9

(dg1)[email protected] [mytest]> alter table table_key1 add key c2_index (c2);

(dg1)[email protected] [mytest]> desc select * from table_key1 where c2 = '3.22';

+----+-------------+------------+------+---------------+----------+---------+-------+------+-------+

| id | select_type | table      | type | possible_keys | key      | key_len | ref   | rows | extra |

|  1 | simple      | table_key1 | ref  | c2_index      | c2_index | 8       | const |    1 | null  |

(dg1)[email protected] [mytest]> alter table table_key1 modify c2 double;

query ok, 0 rows affected (0.03 sec)

|  1 | simple      | table_key1 | ref  | c2_index      | c2_index | 9       | const |    1 | null  |

(dg1)[email protected] [mytest]>

3、看看时间类型

看看date类型的索引长度,在not null和null的时候,分别是3和4

(dg1)[email protected] [mytest]> desc select * from table_key1 where c4 = '2015-05-06';

|  1 | simple      | table_key1 | ref  | c4_index      | c4_index | 3       | const |    4 | using index condition |

row in set, 3 warnings (0.00 sec)

(dg1)[email protected] [mytest]> alter table table_key1 modify c4 date;

query ok, 0 rows affected (0.09 sec)

|  1 | simple      | table_key1 | ref  | c4_index      | c4_index | 4       | const |    4 | using index condition |

在timestamp类型的时候索引长度,在not null 和 null的时候,分别是4和5

(dg1)[email protected] [mytest]> desc select * from table_key1 where c5 = '2015-05-06 11:23:21' ;

|  1 | simple      | table_key1 | ref  | c5_index      | c5_index | 4       | const |    5 | using index condition |

dg1)[email protected] [mytest]> alter table table_key1 modify c5 timestamp on update current_timestamp ;

(dg1)[email protected] [mytest]> desc select * from table_key1 where c5 = '2015-05-06 110:23:21';

|  1 | simple      | table_key1 | ref  | c5_index      | c5_index | 5       | const |    5 | using index condition |

##############################在大家认识里datetime是八个字节的长度,下面就来看看,是不是真的这样

(dg1)root@localhost [mytest]> alter table table_key1 modify c6 datetime not null;

(dg1)root@localhost [mytest]> desc select * from table_key1 where c6 = '2015-05-06 11:10:36';

|  1 | simple      | table_key1 | ref  | c6_index      | c6_index | 5       | const |    1 | null  |

(dg1)root@localhost [mytest]> alter table table_key1 modify c6 datetime  null;

|  1 | simple      | table_key1 | ref  | c6_index      | c6_index | 6       | const |    1 | null  |

颠覆了我们认识,datetime不是8个字节么,下面来看一下mysql的版本,没错mysql5.6是datetime长度是5个字节

(dg1)root@localhost [mytest]> \s

--------------

mysql  ver 14.14 distrib 5.6.22, for linux-glibc2.5 (x86_64) using  editline wrapper

connection id:        3

current database:    mytest

current user:        root@localhost

ssl:            not in use

current pager:        stdout

using outfile:        ''

using delimiter:    ;

server version:        5.6.22-log mysql community server (gpl)

protocol version:    10

connection:        localhost via unix socket

server characterset:    gbk

db     characterset:    gbk

client characterset:    gbk

conn.  characterset:    gbk

unix socket:        /opt/app/mysql/mysql3307.socket

uptime:            4 min 47 sec

threads: 1  questions: 19  slow queries: 0  opens: 75  flush tables: 1  open tables: 64  queries per second avg: 0.066

(dg1)root@localhost [mytest]>

小结:在mysql5.6版本,是否还得使用timestamp类型应该是仁者见仁智者见智的问题了,datetime是五个字节,timestamp范围比较窄(1970-2037年),不排除后续版本会修改其范围值

4.字符类型

表结构,字符集是utf8

看看定长字符类型char的索引长度,在not null 和null中分别为10*3和10*3+1

(dg1)[email protected] [mytest]> alter table table_key add index name_index (name);

query ok, 0 rows affected (0.01 sec)

(dg1)[email protected] [mytest]> explain extended select * from table_key where name='zhangsan';

+----+-------------+-----------+------+---------------+------------+---------+-------+------+----------+-----------------------+

| id | select_type | table     | type | possible_keys | key        | key_len | ref   | rows | filtered | extra                 |

|  1 | simple      | table_key | ref  | name_index    | name_index | 30      | const |    1 |   100.00 | using index condition |

(dg1)[email protected] [mytest]> alter table table_key modify name char(10);

|  1 | simple      | table_key | ref  | name_index    | name_index | 31      | const |    1 |   100.00 | using index condition |

看看变长长字符类型varchar的索引长度,在not null 和null中分别为10*3+2和10*3+2+1

(dg1)[email protected] [mytest]> explain extended select * from table_key where address='shanghai';

+----+-------------+-----------+------+---------------+---------------+---------+-------+------+----------+-----------------------+

| id | select_type | table     | type | possible_keys | key           | key_len | ref   | rows | filtered | extra                 |

|  1 | simple      | table_key | ref  | address_index | address_index | 32      | const |    1 |   100.00 | using index condition |

row in set, 1 warning (0.01 sec)

(dg1)[email protected] [mytest]> alter table table_key modify address varchar(10);

query ok, 0 rows affected (0.10 sec)

|  1 | simple      | table_key | ref  | address_index | address_index | 33      | const |    1 |   100.00 | using index condition |

来看看复合索引的key_len,(刚才测试gbk字符集,字符集转换成gbk了)

| name    | char(10)    | no   |     |         |       |

| address | varchar(10) | no   | mul |         |       |

(dg1)[email protected] [mytest]> alter table table_key drop index name_index;

(dg1)[email protected] [mytest]> alter table table_key drop index address_index;

(dg1)[email protected] [mytest]> alter table table_key add index name_address_index (name,address);

query ok, 0 rows affected (0.02 sec)

(dg1)[email protected] [mytest]> explain extended select * from table_key where address='shanghai' and name='zhangsan';

+----+-------------+-----------+------+--------------------+--------------------+---------+-------------+------+----------+-----------------------+

| id | select_type | table     | type | possible_keys      | key                | key_len | ref         | rows | filtered | extra                 |

|  1 | simple      | table_key | ref  | name_address_index | name_address_index | 42      | const,const |    1 |   100.00 | using index condition |

看看复合索引的长度,因为能全部使用到组合索引,所以是:2*(10)+2*(20)+2=42,下面将name字段允许为空,再来看看

(dg1)[email protected] [mytest]> explain extended select * from table_key where name='zhangsan' and address='shanghai';

|  1 | simple      | table_key | ref  | name_address_index | name_address_index | 43      | const,const |    1 |   100.00 | using index condition |

看看复合索引的长度,因为能全部使用到组合索引,所以是:2*(10)+1+2*(20)+2=43

+----+-------------+-----------+------+--------------------+--------------------+---------+-------+------+----------+-----------------------+

| id | select_type | table     | type | possible_keys      | key                | key_len | ref   | rows | filtered | extra                 |

|  1 | simple      | table_key | ref  | name_address_index | name_address_index | 21      | const |    1 |   100.00 | using index condition |

那么我们来看看部分使用复合索引:2*(10)+1,将address设置为允许为空,再来看看

|  1 | simple      | table_key | ref  | name_address_index | name_address_index | 44      | const,const |    1 |   100.00 | using index condition |

这时候key_len=2*(10)+1+2*(10)+2+1=44

总结

1.整数类型,浮点数类型,时间类型的索引长度

not null=字段本身的字段长度

null=字段本身的字段长度+1,因为需要有是否为空的标记,这个标记需要占用1个字节

datetime类型在5.6中字段长度是5个字节

2.字符类型

varchr(n)变长字段且允许null    =  n * ( utf8=3,gbk=2,latin1=1)+1(null)+2

varchr(n)变长字段且不允许null  =  n * ( utf8=3,gbk=2,latin1=1)+2

char(n)固定字段且允许null      =  n * ( utf8=3,gbk=2,latin1=1)+1(null)

char(n)固定字段且允许null      =  n * ( utf8=3,gbk=2,latin1=1)

变长字段需要额外的2个字节(varchar值保存时只保存需要的字符数,另加一个字节来记录长度(如果列声明的长度超过255,则使用两个字节),所以varcahr索引长度计算时候要加2),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,并且需要额外的存储空间。这个结论在此得到了证实,复合索引有最左前缀的特性,如果复合索引能全部使用上,则是复合索引字段的索引长度之和,这也可以用来判定复合索引是否部分使用,还是全部使用。

<b>本文来自云栖社区合作伙伴“dbgeek”</b>

继续阅读