EXPLAIN 命令详解,b | 鹿晓鹏的个人日常分享

在工作中，我们用于捕捉性能问题最常用的就是打开慢查询，定位执行效率差的SQL，那么当我们定位到一个SQL以后还不算完事，我们还需要知道该SQL的执行计划，比如是全表扫描，还是索引扫描，这些都需要通过EXPLAIN去完成。EXPLAIN命令是查看优化器如何决定执行查询的主要方法。可以帮助我们深入了解MySQL的基于开销的优化器，还可以获得很多可能被优化器考虑到的访问策略的细节，以及当运行SQL语句时哪种策略预计会被优化器采用。需要注意的是，生成的QEP并不确定，它可能会根据很多因素发生改变。MySQL不会将一个QEP和某个给定查询绑定，QEP将由SQL语句每次执行时的实际情况确定，即便使用存储过程也是如此。尽管在存储过程中SQL语句都是预先解析过的，但QEP仍然会在每次调用存储过程的时候才被确定。

通过执行计划可以知道什么？

(root@yayun-mysql-server) [test]>explain select d1.age, t2.id from (select age,name from t1 where id in (1,2))d1, t2 where d1.age=t2.age group by d1.age, t2.id order by t2.id;

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

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

| 1 | PRIMARY | t2 | ref | age | age | 5 | d1.age | 1 | Using where; Using index |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

MySQL执行计划调用方式1.EXPLAIN SELECT ……变体：2.EXPLAIN EXTENDED SELECT ……将执行计划"反编译"成SELECT语句，运行SHOW WARNINGS 可得到被MySQL优化器优化后的查询语句3.EXPLAIN PARTITIONS SELECT ……用于分区表的EXPLAIN生成QEP的信息

执行计划包含的信息

1. id：

包含一组数字，表示查询中执行select子句或操作表的顺序

Example（id相同，执行顺序由上至下）

(root@yayun-mysql-server) [test]>explain select t2.* from t1, t2, t3 where t1.id=t2.id and t1.id=t3.id and t1.name='';

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

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

| 1 | SIMPLE | t1 | ref | PRIMARY,name | name | 63 | const | 1 | Using where; Using index |

| 1 | SIMPLE | t2 | eq_ref | PRIMARY | PRIMARY | 4 | test.t1.id | 1 | |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

Example （如果是子查询，id的序号会递增，id值越大优先级越高，越先被执行）

(root@yayun-mysql-server) [test]>explain select t2.* from t2 where id = (select id from t1 where id = (select t3.id from t3 where t3.name=''));

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

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

| 3 | SUBQUERY | t3 | ref | name | name | 63 | | 1 | Using where; Using index |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

Example（id如果相同，可以认为是一组，从上往下顺序执行；在所有组中，id值越大，优先级越高，越先执行）

(root@yayun-mysql-server) [test]>explain select t2.* from (select t3.id from t3 where t3.name='')s1, t2 where s1.id=t2.id;

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

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

| 1 | PRIMARY | t2 | const | PRIMARY | PRIMARY | 4 | const | 1 | |

| 2 | DERIVED | t3 | ref | name | name | 63 | | 1 | Using where; Using index |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

2. select_type

示查询中每个select子句的类型（简单OR复杂）

a. SIMPLE：查询中不包含子查询或者UNIONb. 查询中若包含任何复杂的子部分，最外层查询则被标记为：PRIMARYc. 在SELECT或WHERE列表中包含了子查询，该子查询被标记为：SUBQUERYd. 在FROM列表中包含的子查询被标记为：DERIVED（衍生）用来表示包含在from子句中的子查询的select，mysql会递归执行并将结果放到一个临时表中。服务器内部称为"派生表"，因为该临时表是从子查询中派生出来的e. 若第二个SELECT出现在UNION之后，则被标记为UNION；若UNION包含在FROM子句的子查询中，外层SELECT将被标记为：DERIVEDf. 从UNION表获取结果的SELECT被标记为：UNION RESULT

SUBQUERY和UNION还可以被标记为DEPENDENT和UNCACHEABLE。DEPENDENT意味着select依赖于外层查询中发现的数据。UNCACHEABLE意味着select中的某些特性阻止结果被缓存于一个item_cache中。

Example

(root@yayun-mysql-server) [test]>explain select d1.name, ( select id from t3) d2 from (select id,name from t1 where name='')d1 union (select name,id from t2);

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

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

| 3 | DERIVED | t1 | ref | name | name | 63 | | 1 | Using where; Using index |

| 2 | SUBQUERY | t3 | index | NULL | age | 5 | NULL | 6 | Using index |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

第一行：id列为1，表示第一个select，select_type列的primary表示该查询为外层查询，table列被标记为<derived3>，表示查询结果来自一个衍生表，其中3代表该查询衍生自第三个select查询，即id为3的select。第二行：id为3，表示该查询的执行次序为2（ 4 => 3），是整个查询中第三个select的一部分。因查询包含在from中，所以为derived。第三行：select列表中的子查询，select_type为subquery，为整个查询中的第二个select。第四行：select_type为union，说明第四个select是union里的第二个select，最先执行。第五行：代表从union的临时表中读取行的阶段，table列的<union1,4>表示用第一个和第四个select的结果进行union操作。

3. type

表示MySQL在表中找到所需行的方式，又称“访问类型”，常见类型如下:

ALL, index, range, ref, eq_ref, const, system, NULL

从左到右，性能从最差到最好

Example

a. ALL：Full Table Scan， MySQL将遍历全表以找到匹配的行

(root@yayun-mysql-server) [test]>explain select * from t1 where email='';

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

b. index：Full Index Scan，index与ALL区别为index类型只遍历索引树

(root@yayun-mysql-server) [test]>explain select id from t1;

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

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

| 1 | SIMPLE | t1 | index | NULL | age | 5 | NULL | 4 | Using index |

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

c. range:索引范围扫描，对索引的扫描开始于某一点，返回匹配值域的行。显而易见的索引范围扫描是带有between或者where子句里带有<, >查询。当mysql使用索引去查找一系列值时，例如IN()和OR列表，也会显示range（范围扫描）,当然性能上面是有差异的。

(root@yayun-mysql-server) [test]>explain select * from t1 where id in (1,4);

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>explain select * from t1 where id between 1 and 4;

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>explain select * from t1 where id=1 or id=4;

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

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

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

row in set (0.01 sec)

(root@yayun-mysql-server) [test]>explain select * from t1 where id > 1;

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

d. ref：使用非唯一索引扫描或者唯一索引的前缀扫描，返回匹配某个单独值的记录行

(root@yayun-mysql-server) [test]>explain select * from t1 where name='yayun';

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

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

| 1 | SIMPLE | t1 | ref | name | name | 63 | const | 1 | Using where |

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

e. eq_ref：类似ref，区别就在使用的索引是唯一索引，对于每个索引键值，表中只有一条记录匹配，简单来说，就是多表连接中使用primary key或者 unique key作为关联条件

(root@yayun-mysql-server) [test]>explain select t1.name from t1, t2 where t1.id=t2.id;

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

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

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

f. const、system：当MySQL对查询某部分进行优化，并转换为一个常量时，使用这些类型访问。如将主键置于where列表中，MySQL就能将该查询转换为一个常量

(root@yayun-mysql-server) [test]>explain select * from ( select * from t1 where id=1)b1;

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

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

| 2 | DERIVED | t1 | const | PRIMARY | PRIMARY | 4 | | 1 | |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

注：system是const类型的特例，当查询的表只有一行的情况下，使用system

g. NULL：MySQL在优化过程中分解语句，执行时甚至不用访问表或索引，例如从一个索引列里选取最小值可以通过单独索引查找完成。

(root@yayun-mysql-server) [test]>explain select * from t1 where id = (select min(id) from t2);

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

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

| 1 | PRIMARY | t1 | const | PRIMARY | PRIMARY | 4 | const | 1 | |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

4. possible_keys指出MySQL能使用哪个索引在表中找到记录，查询涉及到的字段上若存在索引，则该索引将被列出，但不一定被查询使用

5. key显示MySQL在查询中实际使用的索引，若没有使用索引，显示为NULL

Example

(root@yayun-mysql-server) [test]>explain select id,age from t1;

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

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

| 1 | SIMPLE | t1 | index | NULL | age | 5 | NULL | 4 | Using index |

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

6. key_len表示索引中使用的字节数，可通过该列计算查询中使用的索引的长度（key_len显示的值为索引字段的最大可能长度，并非实际使用长度，即key_len是根据表定义计算而得，不是通过表内检索出的）

7. ref表示上述表的连接匹配条件，即哪些列或常量被用于查找索引列上的值

8. rows表示MySQL根据表统计信息及索引选用情况，估算的找到所需的记录所需要读取的行数

Example

(root@yayun-mysql-server) [test]>explain select * from t1 , t2 where t1.id=t2.id and t2.name='atlas';

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

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

| 1 | SIMPLE | t2 | ref | PRIMARY,name | name | 63 | const | 1 | Using where |

| 1 | SIMPLE | t1 | eq_ref | PRIMARY | PRIMARY | 4 | test.t2.id | 1 | |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

9. Extra包含不适合在其他列中显示但十分重要的额外信息a. Using index该值表示相应的select操作中使用了覆盖索引（Covering Index）

Example

(root@yayun-mysql-server) [test]>explain select id from t1;

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

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

| 1 | SIMPLE | t1 | index | NULL | age | 5 | NULL | 4 | Using index |

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

覆盖索引（Covering Index）MySQL可以利用索引返回select列表中的字段，而不必根据索引再次读取数据文件包含所有满足查询需要的数据的索引称为覆盖索引（Covering Index）注意：如果要使用覆盖索引，一定要注意select列表中只取出需要的列，不可select *，因为如果将所有字段一起做索引会导致索引文件过大，查询性能下降

b. Using where表示mysql服务器将在存储引擎检索行后再进行过滤。许多where条件里涉及索引中的列，当（并且如果）它读取索引时，就能被存储引擎检验，因此不是所有带where字句的查询都会显示"Using where"。有时"Using where"的出现就是一个暗示：查询可受益与不同的索引。

Example

(root@yayun-mysql-server) [test]>explain select id,name from t1 where id<4;

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

c. Using temporary表示MySQL需要使用临时表来存储结果集，常见于排序和分组查询

这个值表示使用了内部临时(基于内存的)表。一个查询可能用到多个临时表。有很多原因都会导致MySQL在执行查询期间创建临时表。两个常见的原因是在来自不同表的上使用了DISTINCT,或者使用了不同的ORDER BY和GROUP BY列。可以强制指定一个临时表使用基于磁盘的MyISAM存储引擎。这样做的原因主要有两个：1)内部临时表占用的空间超过min(tmp_table_size，max_heap_table_size)系统变量的限制2)使用了TEXT/BLOB 列

Example

(root@yayun-mysql-server) [test]>explain select id from t1 where id in (1,2) group by age,name;

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

d. Using filesortMySQL中无法利用索引完成的排序操作称为“文件排序”

Example

(root@yayun-mysql-server) [test]>explain select id,age from t1 order by name;

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>explain select id,age from t1 order by age;

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

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

| 1 | SIMPLE | t1 | index | NULL | age | 5 | NULL | 4 | Using index |

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

e. Using join buffer改值强调了在获取连接条件时没有使用索引，并且需要连接缓冲区来存储中间结果。如果出现了这个值，那应该注意，根据查询的具体情况可能需要添加索引来改进能。

Example

(root@yayun-mysql-server) [test]>explain select t1.name from t1 inner join t2 on t1.name=t2.name;

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

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

| 1 | SIMPLE | t2 | ref | name | name | 63 | test.t1.name | 2 | Using where; Using index |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>alter table t1 drop key name;

Query OK, 0 rows affected (0.02 sec)

Records: 0 Duplicates: 0 Warnings: 0

(root@yayun-mysql-server) [test]>alter table t2 drop key name;

Query OK, 0 rows affected (0.02 sec)

Records: 0 Duplicates: 0 Warnings: 0

(root@yayun-mysql-server) [test]>explain select t1.name from t1 inner join t2 on t1.name=t2.name;

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

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

| 1 | SIMPLE | t1 | ALL | NULL | NULL | NULL | NULL | 4 | |

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

rows in set (0.00 sec)

(root@yayun-mysql-server) [test]>

f. Impossible where这个值强调了where语句会导致没有符合条件的行。

Example

(root@yayun-mysql-server) [test]>EXPLAIN SELECT * FROM t1 WHERE 1=2;

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

h. Select tables optimized away这个值意味着仅通过使用索引，优化器可能仅从聚合函数结果中返回一行.

Example

(root@yayun-mysql-server) [test]>explain select max(id) from t1;

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

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

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

row in set (0.00 sec)

(root@yayun-mysql-server) [test]>

I. Index merges当MySQL 决定要在一个给定的表上使用超过一个索引的时候，就会出现以下格式中的一个，详细说明使用的索引以及合并的类型。Using sort_union(...)Using union(...)Using intersect(...)

总结：• EXPLAIN不会告诉你关于触发器、存储过程的信息或用户自定义函数对查询的影响情况• EXPLAIN不考虑各种Cache• EXPLAIN不能显示MySQL在执行查询时所作的优化工作• 部分统计信息是估算的，并非精确值• EXPALIN只能解释SELECT操作，其他操作要重写为SELECT后查看执行计划。