資料量對where in語句的索引影響
阿新 • • 發佈：2019-04-02
我們經常在論壇和麵試中遇到這個問題，mysql中，where in會不會用到索引？
為了徹底搞明白這個問題，做了一些測試，發現記錄數大小對是否命中索引有影響，我們來看一看。
使用的mysql版本是5.7，資料庫引擎為預設的innoDB，索引型別是預設的B+樹索引，用explain執行計劃確認是否命中索引。
我們建立一個表
create table staffs( id int primary key auto_increment, name varchar(24) not null default '' comment '姓名', age int not null default 0 comment '年齡', pos varchar(20) not null default '' comment '職位', add_time timestamp not null default current_timestamp comment '入職時間' )charset utf8 comment '員工記錄表';
1， 我們測試第一種情況，資料量少的情況
先插入三條資料
insert into staffs(name,age,pos,add_time) values('z3',22,'manager',now());
insert into staffs(name,age,pos,add_time) values('July',23,'dev',now());
insert into staffs(name,age,pos,add_time) values('2000',23,'dev',now());1.1 對單列索引的影響，以name為例
alter table staffs add index idx_staffs_name(name);
mysql> explain select * from staffs where name in ('z3', '2000'); +----+-------------+--------+------------+------+-----------------+------+---------+------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+--------+------------+------+-----------------+------+---------+------+------+----------+-------------+ | 1 | SIMPLE | staffs | NULL | ALL | idx_staffs_name | NULL | NULL | NULL | 3 | 66.67 | Using where | +----+-------------+--------+------------+------+-----------------+------+---------+------+------+----------+-------------+ 1 row in set, 1 warning (0.00 sec)
可以看到，沒有命中索引，行數為3，server層對儲存引擎返回的資料做過濾之後剩餘66.67%，也就是說，儲存引擎返回了3條記錄，mysql的server層過濾掉1條，剩下2條，filtered的值為66.67%. （explain詳見之前的博文: https://my.oschina.net/u/3412738/blog/2244825）
1.2 對聯合索引的影響
準備索引
alter table staffs drop index idx_staffs_name;
alter table staffs add index idx_staffs_nameAgePos(name, age, pos);1.2.1 對聯合索引最左欄位的影響
mysql> explain select * from staffs where name = 'z3';
+----+-------------+--------+------------+------+-----------------------+-----------------------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+------+-----------------------+-----------------------+---------+-------+------+----------+-------+
| 1 | SIMPLE | staffs | NULL | ref | idx_staffs_nameAgePos | idx_staffs_nameAgePos | 74 | const | 1 | 100.00 | NULL |
+----+-------------+--------+------------+------+-----------------------+-----------------------+---------+-------+------+----------+-------+
1 row in set, 1 warning (0.00 sec)mysql> explain select * from staffs where name in ('z3', '2000');
+----+-------------+--------+------------+------+-----------------------+------+---------+------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+------+-----------------------+------+---------+------+------+----------+-------------+
| 1 | SIMPLE | staffs | NULL | ALL | idx_staffs_nameAgePos | NULL | NULL | NULL | 3 | 66.67 | Using where |
+----+-------------+--------+------------+------+-----------------------+------+---------+------+------+----------+-------------+
1 row in set, 1 warning (0.04 sec)可以看到，用 = 查詢時，由於最左原則，用到了索引，而用in查詢時，沒有用到索引。
1.2.2 對聯合索引中間欄位的影響
mysql> explain select * from staffs where name = 'z3' and age = 22;
+----+-------------+--------+------------+------+-----------------------+-----------------------+---------+-------------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+------+-----------------------+-----------------------+---------+-------------+------+----------+-------+
| 1 | SIMPLE | staffs | NULL | ref | idx_staffs_nameAgePos | idx_staffs_nameAgePos | 78 | const,const | 1 | 100.00 | NULL |
+----+-------------+--------+------------+------+-----------------------+-----------------------+---------+-------------+------+----------+-------+
1 row in set, 1 warning (0.00 sec)
mysql> explain select * from staffs where name = 'z3' and age in (22, 23);
+----+-------------+--------+------------+------+-----------------------+------+---------+------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+------+-----------------------+------+---------+------+------+----------+-------------+
| 1 | SIMPLE | staffs | NULL | ALL | idx_staffs_nameAgePos | NULL | NULL | NULL | 3 | 66.67 | Using where |
+----+-------------+--------+------------+------+-----------------------+------+---------+------+------+----------+-------------+
1 row in set, 1 warning (0.00 sec)同樣的，當使用 = 查詢時，依次使用了聯合索引，而第二個欄位用 in 查詢時，連第一個欄位都被拖累，沒有使用索引。
2，資料量大的情況
為了快速插入大量資料並建立索引，我們先把原來的那張表drop掉，再建一張一樣的表，不帶任何索引，這樣就不會耗費更新索引的時間。這邊用儲存過程插入。
DELIMITER $$
CREATE PROCEDURE test_insert()
BEGIN
declare i int;
set i = 1 ;
WHILE (i < 10000) DO
INSERT INTO staffs(`name`,`age`,`pos`) VALUES(CONCAT('a', i), FLOOR(20 + RAND() * (100 - i + 1)),'dev');
set i = i + 1;
END WHILE;
commit;
END$$
DELIMITER ;
CALL test_insert();Query OK, 0 rows affected (8 min 7.84 sec)9999條資料耗時8分多鐘，還是有點慢的。
2.1 對單列索引的影響，以name為例
按照之前的動作，建立索引（命令和上面一樣，為了節約篇幅，這裡就不放出來了，下同），再查詢。
mysql> explain select * from staffs where name in ('a1', 'a2000');
+----+-------------+--------+------------+-------+-----------------+-----------------+---------+------+------+----------+-----------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+-------+-----------------+-----------------+---------+------+------+----------+-----------------------+
| 1 | SIMPLE | staffs | NULL | range | idx_staffs_name | idx_staffs_name | 74 | NULL | 2 | 100.00 | Using index condition |
+----+-------------+--------+------------+-------+-----------------+-----------------+---------+------+------+----------+-----------------------+
1 row in set, 1 warning (0.00 sec)命中索引，2條記錄，準確率100%.
1.2 對聯合索引的影響
同樣先刪除單列索引，建立聯合索引。
1.2.1 對聯合索引最左欄位的影響
mysql> explain select * from staffs where name in ('a1', 'a2000');
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| 1 | SIMPLE | staffs | NULL | range | idx_staffs_nameAgePos | idx_staffs_nameAgePos | 74 | NULL | 2 | 100.00 | Using index condition |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
1 row in set, 1 warning (0.00 sec)命中索引。
mysql> explain select * from staffs where name in ('a1', 'a2000') and age = 23;
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| 1 | SIMPLE | staffs | NULL | range | idx_staffs_nameAgePos | idx_staffs_nameAgePos | 78 | NULL | 2 | 100.00 | Using index condition |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
1 row in set, 1 warning (0.00 sec)in欄位後面再加條件也可以命中。
1.2.2 對聯合索引中間欄位的影響
mysql> explain select * from staffs where name = 'a1' and age in (22, 23);
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| 1 | SIMPLE | staffs | NULL | range | idx_staffs_nameAgePos | idx_staffs_nameAgePos | 78 | NULL | 2 | 100.00 | Using index condition |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
1 row in set, 1 warning (0.01 sec)mysql> explain select * from staffs where name in ('a1', 'a2000') and age in (22, 23);
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
| 1 | SIMPLE | staffs | NULL | range | idx_staffs_nameAgePos | idx_staffs_nameAgePos | 78 | NULL | 4 | 100.00 | Using index condition |
+----+-------------+--------+------------+-------+-----------------------+-----------------------+---------+------+------+----------+-----------------------+
1 row in set, 1 warning (0.00 sec)對中間欄位也沒有影響，同樣可以命中索引。
3, 總結
3.1 當資料量少時，會按照聯合索引的順序依次使用索引，反而不會使用單列索引，可能的原因是，mysql認為資料量太小，直接走全表查詢，全表掃描反而更快。
3.2 當資料量大時，單列索引一定會使用。聯合索引也會按順序依次使用。
3.3 當然這裡in條件裡面的數值長度不大，如果是一個很長陣列，導致返回的結果佔全表記錄數量較大時，應該也不會使用索引而走全表查詢。
3.4 這裡還沒有測試，當in條件裡面是一個子查詢時的情況。同時，這裡沒有對5.7以下版本做測試。這裡引用一段這位博主的話
如果是 5.5 之前的版本確實不會走索引的，在 5.5 之後的版本，MySQL 做了優化。MySQL 在 2010 年釋出 5.5 版本中，優化器對 in 操作符可以自動完成優化，針對建立了索引的列可以使用索引，沒有索引的列還是會走全表掃描。
比如，5.5 之前的版本（以下都是 5.5 以前的版本）。select * from a where id in (select id from b); 這條 sql 語句它的執行計劃其實並不是先查詢出 b 表的所有 id，然後再與 a 表的 id 進行比較。mysql 會把 in 子查詢轉換成 exists 相關子查詢，所以它實際等同於這條 sql 語句：select * from a where exists(select * from b where b.id=a.id);
而 exists 相關子查詢的執行原理是：迴圈取出 a 表的每一條記錄與 b 表進行比較，比較的條件是 a.id=b.id。看 a 表的每條記錄的 id 是否在 b 表存在，如果存在就行返回 a 表的這條記