一、使用ROLLUP和CUBE彙總統計

1、測試資料

create table A
(
  GRADE CHAR(5),
  ID    CHAR(5),
  NUM   CHAR(5)
)

insert into a values('a',1,1);

insert into a values('a',2,2);

insert into a values('b',2,4);

insert into a values('b',3,4);

2、檢視幾個分組結果

SQL> select sum(num) from a; //全部求和

  SUM(NUM)
----------
        11

SQL> select grade,sum(num) from a group by grade;//按grade分組

GRADE   SUM(NUM)
----- ----------
a              3
b              8

SQL> select id,sum(num) from a group by id;;//按id分組

ID      SUM(NUM)
----- ----------
1              1
2              6
3              4

SQL> select id,grade,sum(num) from a group by grade,id;;//按grade,id分組

ID    GRADE   SUM(NUM)
----- ----- ----------
1     a              1
2     a              2
2     b              4
3     b              4

3、使用rollup分組結果

SQL> select grade,id,sum(num) from a group by rollup(id,grade);

GRADE ID      SUM(NUM)
----- ----- ----------
a     1              1
      1              1
a     2              2
b     2              4
      2              6
b     3              4
      3              4
                    11

8 rows selected

SQL> select id,grade,sum(num) from a group by rollup(id,grade);

ID    GRADE   SUM(NUM)
----- ----- ----------
1     a              1
1                    1
2     a              2
2     b              4
2                    6
3     b              4
3                    4
                    11

可以看出調換id和grade位置結果是不一樣的，當使用rollup分組時，首先是按第一個引數分組，然後第一個，第二個引數分組 ，如果跟多的話依次類推

4、使用cube分組

SQL> select id,grade,sum(num) from a group by cube(id,grade);

ID    GRADE   SUM(NUM)
----- ----- ----------
                    11
      a              3
      b              8
1                    1
1     a              1
2                    6
2     a              2
2     b              4
3                    4
3     b              4

10 rows selected

SQL> select grade,id,sum(num) from a group by cube(grade,id);

GRADE ID      SUM(NUM)
----- ----- ----------
                    11
      1              1
      2              6
      3              4
a                    3
a     1              1
a     2              2
b                    8
b     2              4
b     3              4

使用cube分組時調換引數的位置並不會影響結果，因為cube會為每個引數單獨分組統計，然後按照然後第一個，第二個引數分組 ，第二個參，第三個引數一次分組統計

5、多個引數實驗

 (1)修改表

alter table a add  grade1 char(5);

SQL> select * from a;

GRADE ID    NUM   GRADE1
----- ----- ----- ------
a     1     1     a1
a     2     2     a1
b     2     4     b1
b     3     4     b2

（2）使用rollup

SQL> select grade,grade1,id,sum(num) from a group by rollup(grade,grade1,id);

GRADE GRADE ID      SUM(NUM)
----- ----- ----- ----------
a     a1    1              1
a     a1    2              2    按grade，grade1和id分組
a     a1                   3     按grade和grade1分組
a                          3         按grade分組
b     b1    2              4
b     b1                   4
b     b2    3              4
b     b2                   4
b                          8
                          11

（2）使用cube

SQL> select grade,grade1,id,sum(num) from a group by cube(grade,grade1,id);

GRADE GRADE ID      SUM(NUM)
----- ----- ----- ----------
                          11
            1              1
            2              6       按id分組
            3              4
      a1                   3      按grade1分組
      a1    1              1     按grade1和id分組
      a1    2              2
      b1                   4
      b1    2              4
      b2                   4
      b2    3              4
a                          3       按grade分組
a           1              1
a           2              2
a     a1                   3     按grade和grade1分組
a     a1    1              1    按grade，id和grade1分組

a     a1    2              2
b                          8
b           2              4      按grade，id分組
b           3              4
b     b1                   4
b     b1    2              4
b     b2                   4
b     b2    3              4

24 rows selected.

其實cube和rollup的區別就是cube採用排列組和，所有的引數排列組合，兒而rollup只是依次組合