mysql sql优化_MySQL SQL优化-让你脑洞大开

由于分库分表的原因，和开发规定了不能使用 表表JOIN 语句。因此，我们要将 JOIN 语句的转化成使用 IN 来做。如现在有 表 A(a_id, c_a)c_a有普通索引,表 B(b_id, c_a) 这两个表要关联, 应该转化为以下步骤处理:

先查询B中的 a_id

SELECT c_a FROM B WHERE xxx;

1

SELECTc_aFROMBWHERExxx;

使用 IN 查询 A 表

SELECT a_id, ... FROM A WHERE c_a IN(在 1 中查出来的 c_a)

1

SELECTa_id,...FROMAWHEREc_aIN(在1中查出来的c_a)

场景

现在表的数据量有 800万。

一般的使用语句是：

SELECT * FROM A WHERE c_a IN(955555, 955556, 955557, 955558, 955559);

1

SELECT *FROMAWHEREc_aIN(955555,955556,955557,955558,955559);

上面语句会执行的很快，知道使用 explain 的都明白这样一般都是会使用索引的，并且是所有范围扫描。

MySQL不会从 1 开始 扫描 800万，而是从555555 扫描到 555559(只要扫描5行数据)。

在一般情况下是没有什么问题的。但是如果 IN 里面的数据是不连续的就有很大问题了。

创建表结构语句

CREATE TABLE t(

id INT unsigned NOT NULL AUTO_INCREMENT,

cid INT unsigned NOT NULL DEFAULT 0,

c1 VARCHAR(50) NOT NULL DEFAULT '',

c2 VARCHAR(50) NOT NULL DEFAULT '',

c3 VARCHAR(50) NOT NULL DEFAULT '',

c4 VARCHAR(50) NOT NULL DEFAULT '',

c5 VARCHAR(50) NOT NULL DEFAULT '',

c6 VARCHAR(50) NOT NULL DEFAULT '',

PRIMARY KEY(id),

INDEX idx$cid(cid)

);

INSERT INTO t VALUES(

NULL,

FLOOR(RAND() * 1000000),

REPEAT('a', 50),

REPEAT('a', 50),

REPEAT('a', 50),

REPEAT('a', 50),

REPEAT('a', 50),

REPEAT('a', 50)

);

-- 重复执行

INSERT INTO t

SELECT NULL,

FLOOR(RAND() * 1000000),

c1,

c2,

c3,

c4,

c5,

c6

FROM t;

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

CREATETABLEt(

idINTunsignedNOTNULLAUTO_INCREMENT,

cidINTunsignedNOTNULLDEFAULT0,

c1VARCHAR(50)NOTNULLDEFAULT'',

c2VARCHAR(50)NOTNULLDEFAULT'',

c3VARCHAR(50)NOTNULLDEFAULT'',

c4VARCHAR(50)NOTNULLDEFAULT'',

c5VARCHAR(50)NOTNULLDEFAULT'',

c6VARCHAR(50)NOTNULLDEFAULT'',

PRIMARYKEY(id),

INDEXidx$cid(cid)

);

INSERTINTOtVALUES(

NULL,

FLOOR(RAND()*1000000),

REPEAT('a',50),

REPEAT('a',50),

REPEAT('a',50),

REPEAT('a',50),

REPEAT('a',50),

REPEAT('a',50)

);

--重复执行

INSERTINTOt

SELECTNULL,

FLOOR(RAND()*1000000),

c1,

c2,

c3,

c4,

c5,

c6

FROMt;

下面是具体的实验过程

使用IN查询连续的数

SELECT *

FROM t

WHERE cid IN(955555, 955556, 955557, 955558, 955559);

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

| id | cid | c1

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

| 319330 | 955555 | aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

| 1885293 | 955555 | aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

| ......

| 8733757 | 955559 | aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

| 8796305 | 955559 | aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

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

41 rows in set (0.15 sec)

1

2

3

4

5

6

7

8

9

10

11

12

13

SELECT *

FROMt

WHEREcidIN(955555,955556,955557,955558,955559);

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

|id|cid|c1

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

|319330|955555|aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

|1885293|955555|aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

|......

|8733757|955559|aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

|8796305|955559|aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

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

41rowsinset(0.15sec)

使用IN查询不连续的数

SELECT *

FROM t

WHERE cid IN(1, 5000, 50000, 500000, 955559);

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

| id | cid | c1

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

| 1 | 341702 | 1 | aaaaaaaaaaaaaaaaaaaaaaaaa

| 1 | 1045176 | 1 | aaaaaaaaaaaaaaaaaaaaaaaaa

......

| 955559 | 8733757 | 955559 | aaaaaaaaaaaaaaaaaaaaaaaaa

| 955559 | 8796305 | 955559 | aaaaaaaaaaaaaaaaaaaaaaaaa

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

41 rows in set (4.34 sec)

1

2

3

4

5

6

7

8

9

10

11

12

13

SELECT *

FROMt

WHEREcidIN(1,5000,50000,500000,955559);

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

|id|cid|c1

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

|1|341702|1|aaaaaaaaaaaaaaaaaaaaaaaaa

|1|1045176|1|aaaaaaaaaaaaaaaaaaaaaaaaa

......

|955559|8733757|955559|aaaaaaaaaaaaaaaaaaaaaaaaa

|955559|8796305|955559|aaaaaaaaaaaaaaaaaaaaaaaaa

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

41rowsinset(4.34sec)

使用UNION优化

SELECT *

FROM (

SELECT 1 AS cid UNION ALL

SELECT 5000 UNION ALL

SELECT 50000 UNION ALL

SELECT 500000 UNION ALL

SELECT 955559

) AS tmp, t

WHERE tmp.cid = t.cid;

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

| id | cid | c1

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

| 1 | 341702 | 1 | aaaaaaaaaaaaaaaaaaaaaaaaa

| 1 | 1045176 | 1 | aaaaaaaaaaaaaaaaaaaaaaaaa

......

| 955559 | 8733757 | 955559 | aaaaaaaaaaaaaaaaaaaaaaaaa

| 955559 | 8796305 | 955559 | aaaaaaaaaaaaaaaaaaaaaaaaa

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

41 rows in set (0.01 sec)

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

SELECT *

FROM(

SELECT1AScidUNIONALL

SELECT5000UNIONALL

SELECT50000UNIONALL

SELECT500000UNIONALL

SELECT955559

)AStmp,t

WHEREtmp.cid=t.cid;

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

|id|cid|c1

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

|1|341702|1|aaaaaaaaaaaaaaaaaaaaaaaaa

|1|1045176|1|aaaaaaaaaaaaaaaaaaaaaaaaa

......

|955559|8733757|955559|aaaaaaaaaaaaaaaaaaaaaaaaa

|955559|8796305|955559|aaaaaaaaaaaaaaaaaaaaaaaaa

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

41rowsinset(0.01sec)

从上面可以看出上面使用UNION的方法生成一个临时表作为关联的主表。

拓展

要是MySQL有只带的一个行转列的函数那就完美了。这样我们就可以不用使用UNION了。

SELECT 1, 5000, 50000, 500000, 955559;

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

| 1 | 5000 | 50000 | 500000 | 955559 |

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

| 1 | 5000 | 50000 | 500000 | 955559 |

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

1 row in set (0.00 sec)

变成以下

SELECT row_to_col(1, 5000, 50000, 500000, 955559);

+--------+

| id |

+--------+

| 1 |

| 5000 |

| 50000 |

| 500000 |

| 955559 |

+--------+

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

SELECT1,5000,50000,500000,955559;

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

|1|5000|50000|500000|955559|

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

|1|5000|50000|500000|955559|

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

1rowinset(0.00sec)

变成以下

SELECTrow_to_col(1,5000,50000,500000,955559);

+--------+

|id|

+--------+

|1|

|5000|

|50000|

|500000|

|955559|

+--------+

要是能像上面就太棒了简直。

感觉本文内容不错，读后有收获？