MySQL EXPLAIN結果集分析 – 附帶大量案例

• 2019 年 10 月 3 日
• 筆記

大量實例助你看懂Explain的輸出內容，輕鬆搞定慢查詢

EXPLAIN：查看SQL語句的執行計劃

EXPLAIN命令可以幫助我們深入了解MySQL基於開銷的優化器，還可以獲得很多可能被優化器考慮到的訪問策略的細節，以及當運行SQL語句時哪種策略預計會被優化器採用，在優化慢查詢時非常有用

執行explain之後結果集包含如下資訊

+----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+-------+  | id | select_type | table | partitions | type | possible_keys | key  | key_len | ref  | rows   | filtered | Extra |  +----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+-------+

下面將對每一個值進行解釋

1、id

id用來標識整個查詢中SELELCT語句的順序，在嵌套查詢中id越大的語句越先執行，該值可能為NULL

id如果相同，從上往下依次執行。id不同，id值越大，執行優先順序越高，如果行引用其他行的並集結果，則該值可以為NULL

2、select_type

select_type表示查詢使用的類型，有下面幾種：

simple： 簡單的select查詢，沒有union或者子查詢

mysql> explain select * from  test where id = 1000;  +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+  | id | select_type | table | partitions | type  | possible_keys | key     | key_len | ref   | rows | filtered | Extra |  +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+  |  1 | SIMPLE      | test  | NULL       | const | PRIMARY       | PRIMARY | 4       | const |    1 |   100.00 | NULL  |  +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+

primary： 最外層的select查詢

mysql> explain select * from (select * from test where id = 1000) a;  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+  | id | select_type | table      | type   | possible_keys | key     | key_len | ref   | rows | Extra |  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+  |  1 | PRIMARY     | <derived2> | system | NULL          | NULL    | NULL    | NULL  |    1 | NULL  |  |  2 | DERIVED     | test       | const  | PRIMARY       | PRIMARY | 8       | const |    1 | NULL  |  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+

union： union中的第二個或隨後的select查詢，不依賴於外部查詢的結果集

mysql> explain select * from test where id = 1000 union all select * from test2 ;  +----+--------------+------------+-------+---------------+---------+---------+-------+-------+-----------------+  | id | select_type  | table      | type  | possible_keys | key     | key_len | ref   | rows  | Extra           |  +----+--------------+------------+-------+---------------+---------+---------+-------+-------+-----------------+  |  1 | PRIMARY      | test       | const | PRIMARY       | PRIMARY | 8       | const |     1 | NULL            |  |  2 | UNION        | test2      | ALL   | NULL          | NULL    | NULL    | NULL  | 67993 | NULL            |  | NULL | UNION RESULT | <union1,2> | ALL   | NULL          | NULL    | NULL    | NULL  |  NULL | Using temporary |  +----+--------------+------------+-------+---------------+---------+---------+-------+-------+-----------------+

dependent union： union中的第二個或隨後的select查詢，依賴於外部查詢的結果集

mysql> explain select * from test where id in (select id  from test where id = 1000 union all select id from test2) ;  +----+--------------------+------------+--------+---------------+---------+---------+-------+-------+-----------------+  | id | select_type        | table      | type   | possible_keys | key     | key_len | ref   | rows  | Extra           |  +----+--------------------+------------+--------+---------------+---------+---------+-------+-------+-----------------+  |  1 | PRIMARY            | test       | ALL    | NULL          | NULL    | NULL    | NULL  | 68505 | Using where     |  |  2 | DEPENDENT SUBQUERY | test       | const  | PRIMARY       | PRIMARY | 8       | const |     1 | Using index     |  |  3 | DEPENDENT UNION    | test2      | eq_ref | PRIMARY       | PRIMARY | 8       | func  |     1 | Using index     |  | NULL | UNION RESULT       | <union2,3> | ALL    | NULL          | NULL    | NULL    | NULL  |  NULL | Using temporary |  +----+--------------------+------------+--------+---------------+---------+---------+-------+-------+-----------------+

subquery： 子查詢中的第一個select查詢，不依賴與外部查詢的結果集

mysql> explain select * from test where id = (select id from test where id = 1000);  +----+-------------+-------+-------+---------------+---------+---------+-------+------+-------------+  | id | select_type | table | type  | possible_keys | key     | key_len | ref   | rows | Extra       |  +----+-------------+-------+-------+---------------+---------+---------+-------+------+-------------+  |  1 | PRIMARY     | test  | const | PRIMARY       | PRIMARY | 8       | const |    1 | NULL        |  |  2 | SUBQUERY    | test  | const | PRIMARY       | PRIMARY | 8       | const |    1 | Using index |  +----+-------------+-------+-------+---------------+---------+---------+-------+------+-------------+

dependent subquery： 子查詢中的第一個select查詢，依賴於外部查詢的結果集

mysql> explain select * from test where id in (select id  from test where id = 1000 union all select id from test2) ;  +----+--------------------+------------+--------+---------------+---------+---------+-------+-------+-----------------+  | id | select_type        | table      | type   | possible_keys | key     | key_len | ref   | rows  | Extra           |  +----+--------------------+------------+--------+---------------+---------+---------+-------+-------+-----------------+  |  1 | PRIMARY            | test       | ALL    | NULL          | NULL    | NULL    | NULL  | 68505 | Using where     |  |  2 | DEPENDENT SUBQUERY | test       | const  | PRIMARY       | PRIMARY | 8       | const |     1 | Using index     |  |  3 | DEPENDENT UNION    | test2      | eq_ref | PRIMARY       | PRIMARY | 8       | func  |     1 | Using index     |  | NULL | UNION RESULT       | <union2,3> | ALL    | NULL          | NULL    | NULL    | NULL  |  NULL | Using temporary |  +----+--------------------+------------+--------+---------------+---------+---------+-------+-------+-----------------+

derived： 用於from子句中有子查詢的情況，mysql會遞歸執行這些子查詢，此結果集放在臨時表中

mysql> explain select * from (select * from test2 where id = 1000)a;  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+  | id | select_type | table      | type   | possible_keys | key     | key_len | ref   | rows | Extra |  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+  |  1 | PRIMARY     | <derived2> | system | NULL          | NULL    | NULL    | NULL  |    1 | NULL  |  |  2 | DERIVED     | test2      | const  | PRIMARY       | PRIMARY | 8       | const |    1 | NULL  |  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+

3、table

table用來表示輸出行所引用的表名

4、type（重要）

type表示訪問類型，下面依次解釋各種類型，類型順序從最好到最差排列

system： 表僅有一行，是const類型的一個特例

mysql> explain select * from (select * from test2 where id = 1000)a;  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+  | id | select_type | table      | type   | possible_keys | key     | key_len | ref   | rows | Extra |  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+  |  1 | PRIMARY     | <derived2> | system | NULL          | NULL    | NULL    | NULL  |    1 | NULL  |  |  2 | DERIVED     | test2      | const  | PRIMARY       | PRIMARY | 8       | const |    1 | NULL  |  +----+-------------+------------+--------+---------------+---------+---------+-------+------+-------+

因為子查詢只有一行數據，模擬了單表只有一行數據，此時type為system

const： 確定只有一行匹配的時候，mysql優化器會在查詢前讀取它並且只讀取一次，速度非常快

mysql> explain select * from test where id =1 ;  +----+-------------+-------+-------+---------------+---------+---------+-------+------+-------+  | id | select_type | table | type  | possible_keys | key     | key_len | ref   | rows | Extra |  +----+-------------+-------+-------+---------------+---------+---------+-------+------+-------+  |  1 | SIMPLE      | test  | const | PRIMARY       | PRIMARY | 8       | const |    1 | NULL  |  +----+-------------+-------+-------+---------------+---------+---------+-------+------+-------+  1 row in set (0.00 sec)

eq_ref： 對於每個來自於前面的表的行組合，從該表中讀取一行，常用在一個索引是unique key或者primary key

mysql> explain select * from test,test2 where test.com_key=test2.com_key;  +----+-------------+-------+--------+---------------+--------------+---------+--------------------+-------+-------+  | id | select_type | table | type   | possible_keys | key          | key_len | ref                | rows  | Extra |  +----+-------------+-------+--------+---------------+--------------+---------+--------------------+-------+-------+  |  1 | SIMPLE      | test2 | ALL    | IDX(com_key)  | NULL         | NULL    | NULL               | 67993 | NULL  |  |  1 | SIMPLE      | test  | eq_ref | IDX(com_key)  | IDX(com_key) | 194     | test.test2.com_key |     1 | NULL  |  +----+-------------+-------+--------+---------------+--------------+---------+--------------------+-------+-------+

ref： 對於來自前面的表的行組合，所有有匹配索引值的行都從這張表中讀取,如果聯接只使用鍵的最左邊的前綴，或如果鍵不是UNIQUE或PRIMARY KEY（換句話說，如果聯接不能基於關鍵字選擇單個行的話），則使用ref

ref可以用於使用=或<=>操作符的帶索引的列

mysql> explain select * from test ,test2 where test.bnet_id=test2.aid;  +----+-------------+-------+------+---------------+---------+---------+-------------------+-------+-----------------------+  | id | select_type | table | type | possible_keys | key     | key_len | ref               | rows  | Extra                 |  +----+-------------+-------+------+---------------+---------+---------+-------------------+-------+-----------------------+  |  1 | SIMPLE      | test  | ALL  | NULL          | NULL    | NULL    | NULL              | 68505 | Using where           |  |  1 | SIMPLE      | test2 | ref  | idx_aid       | idx_aid | 5       | test.test.bnet_id | 34266 | Using index condition |  +----+-------------+-------+------+---------------+---------+---------+-------------------+-------+-----------------------+

test表bnet_id不是索引，test2表aid為索引列

ref_or_null： 類似ref，但是添加了可以專門搜索null值的行

mysql> explain select * from test where bnet_id=1 or bnet_id is null;  +----+-------------+-------+-------------+---------------+----------+---------+-------+------+-----------------------+  | id | select_type | table | type        | possible_keys | key      | key_len | ref   | rows | Extra                 |  +----+-------------+-------+-------------+---------------+----------+---------+-------+------+-----------------------+  |  1 | SIMPLE      | test  | ref_or_null | idx_bnet      | idx_bnet | 9       | const |    2 | Using index condition |  +----+-------------+-------+-------------+---------------+----------+---------+-------+------+-----------------------+

前提為bnet_id列為索引，且bnet_id列有null值

index_merge： 該訪問類型使用了索引合併優化方法，key列包含了使用的索引的清單，key_len包含了使用的索引的最長的關鍵元素

mysql> explain select * from test where id  = 1 or bnet_id = 1;  +----+-------------+-------+-------------+------------------+------------------+---------+------+------+--------------------------------------------+  | id | select_type | table | type        | possible_keys    | key              | key_len | ref  | rows | Extra                                      |  +----+-------------+-------+-------------+------------------+------------------+---------+------+------+--------------------------------------------+  |  1 | SIMPLE      | test  | index_merge | PRIMARY,idx_bnet | PRIMARY,idx_bnet | 8,9     | NULL |    2 | Using union(PRIMARY,idx_bnet); Using where |  +----+-------------+-------+-------------+------------------+------------------+---------+------+------+--------------------------------------------+

前提條件為id列和bnet_id列都有單列索引。如果出現index_merge，並且這類SQL後期使用較頻繁，可以考慮把單列索引換為組合索引，這樣效率更高

range： 只檢索給定範圍的行，使用一個索引來選擇行。key列顯示使用了哪個索引。key_len包含所使用索引的最長關鍵元素。在該類型中ref列為NULL

當使用=、<>、>、>=、<、<=、IS NULL、<=>、BETWEEN或者IN操作符，用常量比較關鍵字列時，可以使用range

mysql> explain select * from test where bnet_id > 1000 and  bnet_id < 10000;  +----+-------------+-------+-------+---------------+----------+---------+------+------+-----------------------+  | id | select_type | table | type  | possible_keys | key      | key_len | ref  | rows | Extra                 |  +----+-------------+-------+-------+---------------+----------+---------+------+------+-----------------------+  |  1 | SIMPLE      | test  | range | idx_bnet      | idx_bnet | 9       | NULL |    1 | Using index condition |  +----+-------------+-------+-------+---------------+----------+---------+------+------+-----------------------+

前提條件為bnet_id列有索引

index： 在進行統計時非常常見，此聯接類型實際上會掃描索引樹

mysql> explain select count(*) from test;  +----+-------------+-------+-------+---------------+----------+---------+------+-------+-------------+  | id | select_type | table | type  | possible_keys | key      | key_len | ref  | rows  | Extra       |  +----+-------------+-------+-------+---------------+----------+---------+------+-------+-------------+  |  1 | SIMPLE      | test  | index | NULL          | idx_bnet | 9       | NULL | 68505 | Using index |  +----+-------------+-------+-------+---------------+----------+---------+------+-------+-------------+

all： 對於每個來自於先前的表的行組合，進行完整的表掃描，通常可以增加更多的索引而不要使用ALL，使得行能基於前面的表中的常數值或列值被檢索出

mysql> explain select *  from test where create_time = '0000-00-00 00:00:00';  +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+  | id | select_type | table | type | possible_keys | key  | key_len | ref  | rows  | Extra       |  +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+  |  1 | SIMPLE      | test  | ALL  | NULL          | NULL | NULL    | NULL | 68505 | Using where |  +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+

5、possible_keys

possible_keys是指在這個SQL中，mysql可以使用這個索引去輔助查找記錄，當查詢涉及到的欄位，都會被列出，但不一定被查詢使用.若為空則表示沒有可以使用的索引，此時可以通過檢查where語句看是否可以引用某些列或者新建索引來提高性能。

6、key（重要）

key列顯示的是當前表實際使用的索引，如果沒有選擇索引，則此列為null，要想強制MySQL使用或忽視possible_keys列中的索引，在查詢中使用FORCE INDEX、USE INDEX或者IGNORE INDEX

7、key_len

key_len列顯示MySQL決定使用的鍵長度。如果KEY鍵是NULL，則長度為NULL。在不損失精確性的情況下，長度越短越好

key len的長度還和字符集有關,latin1一個字元佔用1個位元組,gbk一個字元佔用2個位元組,utf8一個字元佔用3個位元組。key_len的計演算法方法：

key_len只指示了where中用於條件過濾時被選中的索引列，是不包含order by或group by這一部分被選中的索引列

8、ref

ref列用來顯示使用哪個列或常數與key一起從表中選擇相應的行。它顯示的列的名字（或const），此列多數時候為null

9、rows

rows列顯示的是mysql解析器認為執行此SQL時必須掃描的行數。此數值為一個預估值，不是具體值，通常比實際值小

10、filtered

此參數為mysql 5.7 新加參數，指的是返回結果的行數所佔需要讀到的行（rows的值）的比例
對於使用join時，前一個表的結果集大小直接影響了循環的行數

11、extra（重要）

extra表示不在其他列並且也很重要的額外資訊

using index： 該值表示這個SQL語句使用了覆蓋索引（覆蓋索引是指可以直接在索引列中得到想要的結果，而不用去回表），此時效率最高

mysql> explain select id from test;  +----+-------------+-------+-------+---------------+----------+---------+------+-------+-------------+  | id | select_type | table | type  | possible_keys | key      | key_len | ref  | rows  | Extra       |  +----+-------------+-------+-------+---------------+----------+---------+------+-------+-------------+  |  1 | SIMPLE      | test  | index | NULL          | idx_bnet | 9       | NULL | 68505 | Using index |  +----+-------------+-------+-------+---------------+----------+---------+------+-------+-------------+

這個例子中id欄位為主鍵，但是key那裡顯示走的並不是主鍵索引，這個是因為mysql的所有二級索引中都會包含所有的主鍵資訊，而mysql沒有單獨的存儲主鍵索引，所以掃描二級索引的開銷比全表掃描更快

using where： 表示存儲引擎搜到記錄後進行了後過濾(POST-FILTER)，如果查詢未能使用索引，using where的作用只是提醒我們mysql要用where條件過濾結果集

mysql> explain select * from test where id > 1;  +----+-------------+-------+-------+---------------+---------+---------+------+-------+-------------+  | id | select_type | table | type  | possible_keys | key     | key_len | ref  | rows  | Extra       |  +----+-------------+-------+-------+---------------+---------+---------+------+-------+-------------+  |  1 | SIMPLE      | test  | range | PRIMARY       | PRIMARY | 8       | NULL | 34252 | Using where |  +----+-------------+-------+-------+---------------+---------+---------+------+-------+-------------+

using temporary 表示mysql需要使用臨時表來存儲結果集，常見於排序和分組查詢

mysql> explain select * from test where id in (1,2) group by bnet_id;  +----+-------------+-------+-------+-----------------------------------------+---------+---------+------+------+----------------------------------------------+  | id | select_type | table | type  | possible_keys                           | key     | key_len | ref  | rows | Extra                                        |  +----+-------------+-------+-------+-----------------------------------------+---------+---------+------+------+----------------------------------------------+  |  1 | SIMPLE      | test  | range | PRIMARY,IDX(event_key-bnet_Id),idx_bnet | PRIMARY | 8       | NULL |    2 | Using where; Using temporary; Using filesort |  +----+-------------+-------+-------+-----------------------------------------+---------+---------+------+------+----------------------------------------------+

using filesort： 是指mysql無法利用索引直接完成排序（排序的欄位不是索引欄位），此時會用到緩衝空間來進行排序

mysql> explain select * from test order by bnet_id;  +----+-------------+-------+------+---------------+------+---------+------+-------+----------------+  | id | select_type | table | type | possible_keys | key  | key_len | ref  | rows  | Extra          |  +----+-------------+-------+------+---------------+------+---------+------+-------+----------------+  |  1 | SIMPLE      | test  | ALL  | NULL          | NULL | NULL    | NULL | 68505 | Using filesort |  +----+-------------+-------+------+---------------+------+---------+------+-------+----------------+

using join buffer： 強調在獲取連接條件時沒有用到索引，並且需要連接緩衝區來存儲中間結果。（性能可以通過添加索引或者修改連接欄位改進）

mysql> explain select * from test left join test2 on test.create_time = test2.create_time;  +----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+----------------------------------------------------+  | id | select_type | table | partitions | type | possible_keys | key  | key_len | ref  | rows   | filtered | Extra                                              |  +----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+----------------------------------------------------+  |  1 | SIMPLE      | test  | NULL       | ALL  | NULL          | NULL | NULL    | NULL | 959692 |   100.00 | NULL                                               |  |  1 | SIMPLE      | test2 | NULL       | ALL  | NULL          | NULL | NULL    | NULL | 958353 |   100.00 | Using where; Using join buffer (Block Nested Loop) |  +----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+----------------------------------------------------+  2 rows in set, 1 warning (0.00 sec)

Block Nested Loop是指Block Nested-Loop Join演算法：將外層循環的行/結果集存入join buffer, 內層循環的每一行與整個buffer中的記錄做比較，從而減少內層循環的次數.

impossible where： 表示where條件導致沒有返回的行

mysql> explain select * from test where id is null;  +----+-------------+-------+------+---------------+------+---------+------+------+------------------+  | id | select_type | table | type | possible_keys | key  | key_len | ref  | rows | Extra            |  +----+-------------+-------+------+---------------+------+---------+------+------+------------------+  |  1 | SIMPLE      | NULL  | NULL | NULL          | NULL | NULL    | NULL | NULL | Impossible WHERE |  +----+-------------+-------+------+---------------+------+---------+------+------+------------------+

using index condition： 是mysql 5.6 之後新加的特性，結合mysql的ICP（Index Condition Pushdown）特性使用。主要是優化了可以在索引（僅限二級索引）上進行 like 查找

如果extra中出現多個上面結果，則表示順序使用上面的方法進行解析查詢

相關文章推薦閱讀：

• 中小團隊快速構建SQL自動審核系統
• 運維效率之數據遷移自動化