在优化SQL语句中使用虚拟索引
定义:虚拟索引(virtual index) 是指没有创建对应的物理段的索引。
虚拟索引的目的:是在不损耗主机CPU,IO,磁盘空间去实际创建索引的情况下,来判断一个索引是否能够对SQL优化起到作用。列如我们在优化一条SQL语句的时候,通常会查看需要优化的语句的执行计划,在考虑是否需要在表的某列上建立索引时就可以用到虚拟索引。虚拟索引建立的时候因为其没有消耗主机的相关资源,因此可以在相当快的时间内建立完成。
下面我们来看一下试验:
首先建立两张测试表
create table bigtab as select rownum as id,a.* from sys.all_objects a;
create table smalltab as select rownum as id,a.* from a;
create table smalltab as select rownum as id,a.* from sys.all_tables a;
多次运行以下语句,以插入多一些测试数据:
insert into bigtab select ronum as id,a.* from sys.all_objects a;
insert into smalltab select rownum as id,a.* from sys.all_tables a;
查看需要执行语句的执行计划:
SQL> explain plan for select count(*) from bigtab a,smalltab b where a.object_name=b.table_name; Explained. SQL> select * from table(dbms_xplan.display()); PLAN_TABLE_OUTPUT -------------------------------------------------------------------------------- Plan hash value: 3089226980 -------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | -------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 1 | 40 | 518 (1)| 00:00:07 | | 1 | SORT AGGREGATE | | 1 | 40 | | | |* 2 | HASH JOIN | | 99838 | 3899K| 518 (1)| 00:00:07 | | 3 | TABLE ACCESS FULL| SMALLTAB | 15311 | 299K| 172 (0)| 00:00:03 | | 4 | TABLE ACCESS FULL| BIGTAB | 85284 | 1665K| 345 (1)| 00:00:05 | -------------------------------------------------------------------------------- PLAN_TABLE_OUTPUT -------------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 2 - access("A"."OBJECT_NAME"="B"."TABLE_NAME") 16 rows selected.
下面我们在两个表上创建两个虚拟索引,分别在object_name和table_name列上,看看优化器是否会使用这两个索引,以及优化器的成本会如何变化。
SQL> show parameter _use_nosegment SQL> alter session set "_use_nosegment_indexes"=true; Session altered. SQL> show parameter _use_nosegment NAME TYPE VALUE ------------------------------------ ----------- ------------------------------ _use_nosegment_indexes boolean TRUE SQL> create index big_ind on bigtab(object_name) nosegment; Index created. SQL> create index small_ind on smalltab(table_name) nosegment;
create index语句的NOSEGMENT选项表明这个索引是“虚拟的”-----就是没有与之相关联的实际索引段。如果优化器认为这个索引有用,参数_use_nosegment_indexes将指示数据库可以在执行计划中使用这些索引。下面我们来看看如果真的创建这些索引,它们是否值得使用:
SQL> explain plan for select count(*) from bigtab a,smalltab b where a.object_name=b.table_name; Explained. SQL> select * from table(dbms_xplan.display()); PLAN_TABLE_OUTPUT -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Plan hash value: 1254475829 ------------------------------------------------------------------------------------ | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | ------------------------------------------------------------------------------------ | 0 | SELECT STATEMENT | | 1 | 40 | 19 (6)| 00:00:01 | | 1 | SORT AGGREGATE | | 1 | 40 | | | |* 2 | HASH JOIN | | 99838 | 3899K| 19 (6)| 00:00:01 | | 3 | INDEX FAST FULL SCAN| SMALL_IND | 15311 | 299K| 9 (0)| 00:00:01 | | 4 | INDEX FAST FULL SCAN| BIG_IND | 85284 | 1665K| 9 (0)| 00:00:01 | ------------------------------------------------------------------------------------ PLAN_TABLE_OUTPUT -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 2 - access("A"."OBJECT_NAME"="B"."TABLE_NAME") 16 rows selected.
执行计划显示,这两个索引确实被认为有用,优化器估算的成本也下降到19了。所以如果需要优化语句可以建议创建这些索引。