Oracle TABLE ACCESS BY INDEX ROWID 说明
一. 测试环境
SQL> select * from v$version where rownum=1;
BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release11.2.0.3.0 - 64bit Production
SQL> create table dave as selectobject_id,object_name,object_type,created,timestamp,status from all_objects;
表已创建。
SQL> create table dave2 as select * from dave;
表已创建。
--收集统计信息,这里没有收集直方图:
SQL> exec dbms_stats.gather_table_stats(ownname=>'SYS',tabname =>'DAVE',estimate_percent => 10 ,method_opt =>'FORCOLUMNS size 1',degree=>10,cascade => true);
PL/SQL 过程已成功完成。
SQL> exec dbms_stats.gather_table_stats(ownname=>'SYS',tabname =>'DAVE2',estimate_percent => 10 ,method_opt =>'FORCOLUMNS size 1',degree=>10,cascade => true);
PL/SQL 过程已成功完成。
--避免其他影响,先刷新buffer cache:
SQL> alter system flush buffer_cache;
系统已更改。
--查看全表扫描时的执行计划:
SQL> set autot traceonly
SQL> select d1.object_name,d2.object_type fromdave d1,dave2 d2 where d1.object_id=d2.object_id;
已选择72762行。
执行计划
----------------------------------------------------------
Plan hash value: 3613449503
------------------------------------------------------------------------------------
| Id |Operation | Name | Rows | Bytes |TempSpc| Cost (%CPU)| Time |
------------------------------------------------------------------------------------
| 0 |SELECT STATEMENT | | 72520 | 3824K| | 695 (1)| 00:00:09 |
|* 1 | HASH JOIN | | 72520 | 3824K| 2536K| 695 (1)| 00:00:09 |
| 2 | TABLE ACCESS FULL| DAVE2 | 71990 | 1687K| | 213 (1)| 00:00:03 |
| 3 | TABLE ACCESS FULL| DAVE | 72520 | 2124K| | 213 (1)| 00:00:03 |
------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 -access("D1"."OBJECT_ID"="D2"."OBJECT_ID")
统计信息
----------------------------------------------------------
0 recursive calls
0 db block gets
6353 consistent gets
1558 physical reads
0 redo size
3388939 bytes sent via SQL*Net toclient
53874 bytes received via SQL*Netfrom client
4852 SQL*Net roundtrips to/fromclient
0 sorts (memory)
0 sorts (disk)
72762 rows processed
--这里产生了1558的物理读
SQL>
--在object_id上创建索引:
SQL> create index idx_dave_object_idon dave(object_id);
索引已创建。
SQL> create index idx_dave_object_id2 ondave2(object_id);
索引已创建。
--在次查看执行计划:
SQL> select d1.object_name,d2.object_type fromdave d1,dave2 d2 where d1.object_id=d2.object_id;
已选择72762行。
执行计划
----------------------------------------------------------
Plan hash value: 3613449503
------------------------------------------------------------------------------------
| Id |Operation | Name | Rows | Bytes |TempSpc| Cost (%CPU)| Time |
------------------------------------------------------------------------------------
| 0 |SELECT STATEMENT | | 72520 | 3824K| | 695 (1)| 00:00:09 |
|* 1 | HASH JOIN | | 72520 | 3824K| 2536K| 695 (1)| 00:00:09 |
| 2 | TABLE ACCESS FULL| DAVE2 | 71990 | 1687K| | 213 (1)| 00:00:03 |
| 3 | TABLE ACCESS FULL| DAVE | 72520 | 2124K| | 213 (1)| 00:00:03 |
------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 -access("D1"."OBJECT_ID"="D2"."OBJECT_ID")
统计信息
----------------------------------------------------------
1 recursive calls
0 db block gets
6353 consistent gets
0 physical reads
0 redo size
3388939 bytes sent via SQL*Net toclient
53874 bytes received via SQL*Netfrom client
4852 SQL*Net roundtrips to/fromclient
0 sorts (memory)
0 sorts (disk)
72762 rows processed
这里的物理读为0. 但是还是走的是全表扫描。
--刷新一下buffer,增加索引条件:
SQL> alter system flush buffer_cache;
系统已更改。
SQL> select d1.object_name,d2.object_type fromdave d1,dave2 d2 where d1.object_id=d2.object_id and d1.object_id <100;
已选择98行。
执行计划
----------------------------------------------------------
Plan hash value: 504164237
----------------------------------------------------------------------------------------------------
| Id |Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
----------------------------------------------------------------------------------------------------
| 0 |SELECT STATEMENT | | 3600 | 189K| 23 (5)| 00:00:01 |
|* 1 | HASH JOIN | | 3600 | 189K| 23 (5)| 00:00:01 |
| 2 | TABLE ACCESS BY INDEX ROWID| DAVE2 | 3600 | 86400 | 11 (0)| 00:00:01 |
|* 3 | INDEX RANGE SCAN | IDX_DAVE_OBJECT_ID2 | 648 | | 3 (0)| 00:00:01 |
| 4 | TABLE ACCESS BY INDEX ROWID| DAVE | 3626 | 106K| 11 (0)| 00:00:01 |
|* 5 | INDEX RANGE SCAN | IDX_DAVE_OBJECT_ID | 653| | 3 (0)| 00:00:01 |
----------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 -access("D1"."OBJECT_ID"="D2"."OBJECT_ID")
3 -access("D2"."OBJECT_ID"<100)
5 -access("D1"."OBJECT_ID"<100)
统计信息
----------------------------------------------------------
1 recursive calls
0 db block gets
20 consistent gets
6 physical reads
0 redo size
3317 bytes sent via SQL*Net toclient
590 bytes received via SQL*Netfrom client
8 SQL*Net roundtrips to/fromclient
0 sorts (memory)
0 sorts (disk)
98 rows processed
SQL>
走索引之后,物理读从1558降到6.
二.说明
在上面的测试中,我们看到了索引扫描的类型和多表关联的类型,关于这几种类型的说明,参考:
http://blog.csdn.net/tianlesoftware/article/details/5852106
多表连接的三种方式详解 HASH JOIN MERGE JOINNESTED LOOP
http://blog.csdn.net/tianlesoftware/article/details/5826546
从执行计划中,当我们走索引之后,在对应的表上就会出现:
TABLE ACCESS BY INDEX ROWID
在如下文章中对OracleROWID 有说明。
http://blog.csdn.net/tianlesoftware/article/details/5020718
rowid是伪列(pseudocolumn),在查询结果输出时它被构造出来的。rowid并不会真正存在于表的data block中,其存在于index当中,用来通过rowid来寻找表中的行数据。
ROWID 由以下几部分组成:
1. 数据对象编号:每个数据对象(如表或索引)在创建时都分配有此编号,并且此编号在数据库中是唯一的
2. 相关文件编号:此编号对于表空间中的每个数据文件是唯一的
3. 块编号:表示包含此行的块在数据文件中的位置
4. 行编号:标识块头中行目录位置的位置
Oracle 索引中保存的是我们字段的值和该值对应的rowid,我们根据索引进行查找时,就会返回该block的rowid,然后根据rowid直接去block上去我们需要的数据,因此就出现了:
因为ROWID 对应一个block,所以当使用TABLE ACCESS BY INDEX ROWID时,每次就只能读取一个block。
假设我们我们的数据返回100个ROWID,其中10个row 位于同一个block上,那么我们只需要访问91次block,就可以拿到我们需要的数据。
关于如何确定row记录在哪个block的方法参考:
http://blog.csdn.net/tianlesoftware/article/details/6529346
小结:
(1) TABLE ACCESS BY INDEX ROWID 只出现在使用索引的情况下。
(2) TABLE ACCESS BY INDEX ROWID 是单块读,每次只能读取一个block。
-------------------------------------------------------------------------------------------------------
版权所有,文章允许转载,但必须以链接方式注明源地址,否则追究法律责任!
Skype: tianlesoftware
QQ: tianlesoftware@gmail.com
Email: tianlesoftware@gmail.com
Blog: http://www.tianlesoftware.com
Weibo: http://weibo.com/tianlesoftware
Twitter: http://twitter.com/tianlesoftware
Facebook: http://www.facebook.com/tianlesoftware
Linkedin: http://cn.linkedin.com/in/tianlesoftware
-------加群需要在备注说明Oracle表空间和数据文件的关系,否则拒绝申请----
DBA1 群:62697716(满); DBA2 群:62697977(满) DBA3 群:62697850(满)
DBA 超级群:63306533(满); DBA4 群:83829929 DBA5群: 142216823
DBA6 群:158654907 DBA7 群:172855474 DBA总群:104207940