如何降低索引的clustering_factor

如何降低索引的clustering_factor 2012-06-08 22:40:19

分类： Oracle

环境:
OS:Red Hat Linux As 5
DB:10.2.0.4

我们知道判断一个索引的好坏可以通过该索引的clustering_factor高低来衡量,clustering_factor越低,索引的使用效果就越好,那怎么样才能降低索引的clustering_factor呢,通常使用的方法是让表的索引字段值按顺序存储,下面通过一个例子说明.

1.创建表并构造无序的数据
create table scott.tb_index_test
(
id number not null,
name varchar2(30)
);

create index scott.idx_tb_index_test on scott.tb_index_test(id);

declare
l_random_value number;
begin
for i in 1 .. 100000 loop
l_random_value := round(dbms_random.value(1,100000));
insert into scott.tb_index_test(id) values (l_random_value);
commit;
end loop;
end;

2.分析表

begin
dbms_stats.gather_table_stats(ownname => 'SCOTT',
tabname => 'TB_INDEX_TEST',
cascade => true);
end;

3.查看索引当前的clustering_factor

SQL> select ui.clustering_factor, ui.num_rows, ui.index_type, ui.distinct_keys
from dba_indexes ui where ui.table_name = 'TB_INDEX_TEST';
CLUSTERING_FACTOR NUM_ROWS INDEX_TYPE DISTINCT_KEYS
----------------- ---------- --------------------------- -------------
99742 100000 NORMAL 100000

4.将表中的数据按照索引字段存储

create table scott.tmp as select * from scott.tb_index_test;
truncate table scott.tb_index_test;
insert into scott.tb_index_test select * from scott.tmp order by id;

5.再次分析表

begin
dbms_stats.gather_table_stats(ownname => 'SCOTT',
tabname => 'TB_INDEX_TEST',
cascade => true);
end;

6.这个时候索引的clustering_factor明显降低了

SQL> select ui.clustering_factor, ui.num_rows, ui.index_type, ui.distinct_keys
from dba_indexes ui where ui.table_name = 'TB_INDEX_TEST';
CLUSTERING_FACTOR NUM_ROWS INDEX_TYPE DISTINCT_KEYS
----------------- ---------- --------------------------- -------------
372 100000 NORMAL 100000

-- The End --