创建一个表时表中列的顺序在某些程度上对性能会有一定的影响。(表中的列有数据) 。
Oracle对行数据的存储结构ROW HEADER(行头)和COLUMN DATA(列数据)。ROW HEADER存储的信息是一个FLAG BYTE，一个LOCK BYTE 和COLUMN，COUNT.COLUMN DATA包含COLUMN LENGTH和COLUMN DATA。
关于这些我们可以DUMP个表做一下测试：
create test as select * from dba_objects;
select header_file,header_block from
dba_segments where owner='TEST' and segment_name='TEST';
HEADER_FILEHEADER_BLOCK
------------------------ ---------------------------
13 1179
alter system dump datafile 13 block 1180 |
得出来的文件在UDUMP中。
我们查看如下信息：
block_row_dump:
tab 0, row 0, @0x1f20
tl: 96 fb: --H-FL-- lb: 0x0 cc: 13--------row header信息.
col 0: [ 3] 53 59 53-------------------COLUMN DATA
fb: --H-FL-- 是FLAG BYTE.
fb Flag Byte:
K = Cluster Key (Flags may change meaning if this is set to show HASH cluster)
C = Cluster table member
H = Head piece of row
D = Deleted row
F = First data piece
L = Last data piece
P = First column continues from previous piece
N = Last column continues in next piece
lb: 0x0-----------LOCK BYTE,锁信息
cc: 13------------COLUMN COUNT
col 0---------第一列
[ 3]-------------COLUMN LENGTH
53 59 53---------实际数据 |
列的值,Oracle首先做的是检查这些相关列的长度位.这个操作比较快而且效率较高。但是如果反复频繁的这样子做还是会带来性能方面的影响。
下面的例子中创建了一个有10列的表并插入数据。
先设置DB_BLOCK_SIZE=2K(用参数设置,在这里设置为这个只是为了测试方便)：
SQL> create table small (
2n0 number,
3n1 number,
4n2 number,
5n3 number,
6n4 number,
7n5 number,
8n6 number,
9n7 number,
10n8 number,
11n9 number
12 ) pctfree 0;
Table created.
SQL> begin
2for i in 1..78 loop
3 insert into small values (0,0,0,0,0,0,0,0,0,0);
4end loop;
5 end;
6 /
PL/SQL procedure successfully completed.
SQL> set timing on
SQL> declare
2n number;
3 begin
4for i in 1..1000000 loop
5 select sum(n0) into n from small;
6end loop;
7 end;
8 /
PL/SQL procedure successfully completed.
Elapsed: 00:07:437.30
SQL> declare
2n number;
3 begin
4for i in 1..1000000 loop
5 select sum(n9) into n from small;
6end loop;
7 end;
8 /
PL/SQL procedure successfully completed.
Elapsed: 00:08:482.13 |
