这两天听了将近20场演讲,感觉收获很多,最深的感觉就是自己还有很长的路要走。有几个点记录一下:
昨天听老猫讲,提到一个普遍的问题就是Oracle里count(*)、count(1)和count(主键)到底哪个快的问题。这个问题看起来很简单,每个人都会有自己的答案,去百度上搜会出来一大堆帖子来讲哪个更快。但是老猫说了它们三个其实是一样的,我听到之后也觉得挺诧异的,因为我记得别人跟我说过count(主键)会快,然后自己简单想了一下,觉得好像是那么回事的就没有深入去追究。接着老猫说官方有这样的说法这三个其实是等价的。晚上回来之后到MOS上查了一下,居然被我找到了How the Oracle CBO Chooses a Path for the SELECT COUNT(*) Command (文档 ID 124717.1)。这篇文档讲的就是在CBO优化器模式下,Oracle怎样去评估没有where条件select count(*)和select count(colum)语句的最优路径。
1、创建测试表并设计测试场景:
--创建测试表
sys@ORCL>create table journal_entries
2 (id_je number(8) ,
3 date_je date not null,
4 balanced number ,
5 constraint indx_ecr_id_je primary key(id_je)
6 );
Table created.
--创建索引
sys@ORCL>create index indx_ecr_date_je_balanced on journal_entries(date_je,balanced);
Index created.
sys@ORCL>create index indx_ecr_balanced_date_je on journal_entries(balanced,date_je);
Index created.
sys@ORCL>create index indx_ecr_balanced on journal_entries(balanced);
Index created.
--插入测试数据
sys@ORCL>insert into journal_entries values(1,sysdate,11);
1 row created.
sys@ORCL>insert into journal_entries values(2,sysdate,21);
1 row created.
sys@ORCL>insert into journal_entries values(3,sysdate,31);
1 row created.
sys@ORCL>insert into journal_entries values(4,sysdate,41);
1 row created.
sys@ORCL>insert into journal_entries values(5,sysdate,51);
1 row created.
sys@ORCL>insert into journal_entries values(6,sysdate,61);
1 row created.
sys@ORCL>insert into journal_entries values(7,sysdate,71);
1 row created.
sys@ORCL>insert into journal_entries values(8,sysdate,81);
1 row created.
sys@ORCL>insert into journal_entries values(9,sysdate,91);
1 row created.
sys@ORCL>commit;
Commit complete.
--收集统计信息
sys@ORCL>exec dbms_stats.gather_table_stats(ownname=>USER,tabname=>'JOURNAL_ENTRIES',cascade=>true);
PL/SQL procedure successfully completed.
设计四个场景进行对比:
Sel2 : Select count(1) from journal_entries;
Sel3 : Select count(id_je) from journal_entries;
Sel4 : Select count(balanced) from journal_entries;
1、场景1和场景2等价
For CBO, Sel1 and Sel2 are strictly equivalent
sys@ORCL>alter session set statistics_level=all;
Session altered.
sys@ORCL>select count(*) from journal_entries;
COUNT(*)
----------
9
sys@ORCL>select * from table(dbms_xplan.display_cursor(null,null,'runstats_last'));
PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
SQL_ID 5ja3ukp4wd73p, child number 0
-------------------------------------
select count(*) from journal_entries
Plan hash value: 42135099
---------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows | A-Rows | A-Time | Buffers |
---------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | 1 |00:00:00.01 | 1 |
| 1 | SORT AGGREGATE | | 1 | 1 | 1 |00:00:00.01 | 1 |
| 2 | INDEX FULL SCAN| INDX_ECR_ID_JE | 1 | 9 | 9 |00:00:00.01 | 1 |
---------------------------------------------------------------------------------------------
14 rows selected.
sys@ORCL>select count(1) from journal_entries;
COUNT(1)
----------
9
sys@ORCL>select * from table(dbms_xplan.display_cursor(null,null,'runstats_last'));
PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
SQL_ID gbxjjuqj9j7ww, child number 0
-------------------------------------
select count(1) from journal_entries
Plan hash value: 42135099
---------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows | A-Rows | A-Time | Buffers |
---------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | 1 |00:00:00.01 | 1 |
| 1 | SORT AGGREGATE | | 1 | 1 | 1 |00:00:00.01 | 1 |
| 2 | INDEX FULL SCAN| INDX_ECR_ID_JE | 1 | 9 | 9 |00:00:00.01 | 1 |
---------------------------------------------------------------------------------------------
14 rows selected.
可以看到两个语句的执行计划是完全相同的。
2、场景3也与前两个场景等价,因为id_je有NOT NULL约束
For Sel3, CBO does the same as for Sel1 and Sel2 since "id_je" has a NOT NULL constraint.
sys@ORCL>select count(id_je) from journal_entries;
COUNT(ID_JE)
------------
9
sys@ORCL>select * from table(dbms_xplan.display_cursor(null,null,'runstats_last'));
PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
SQL_ID b1p4v15dwx7hs, child number 0
-------------------------------------
select count(id_je) from journal_entries
Plan hash value: 42135099
---------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows | A-Rows | A-Time | Buffers |
---------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | 1 |00:00:00.01 | 1 |
| 1 | SORT AGGREGATE | | 1 | 1 | 1 |00:00:00.01 | 1 |
| 2 | INDEX FULL SCAN| INDX_ECR_ID_JE | 1 | 9 | 9 |00:00:00.01 | 1 |
---------------------------------------------------------------------------------------------
14 rows selected.
可以看到执行计划与前两个也是完全相同的。
4、场景4跟前边3个不同,因为balanced列上没有NOT NULL约束,但是balanced列上有索引,那会走这个列上的索引么?我们来看一下执行计划:
sys@ORCL>select count(balanced) from journal_entries;
COUNT(BALANCED)
---------------
9
sys@ORCL>select * from table(dbms_xplan.display_cursor(null,null,'runstats_last'));
PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
SQL_ID bc3bc8c0fg14z, child number 0
-------------------------------------
select count(balanced) from journal_entries
Plan hash value: 3638043346
--------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows | A-Rows | A-Time | Buffers |
--------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | 1 |00:00:00.01 | 1 |
| 1 | SORT AGGREGATE | | 1 | 1 | 1 |00:00:00.01 | 1 |
| 2 | INDEX FULL SCAN| INDX_ECR_DATE_JE_BALANCED | 1 | 9 | 9 |00:00:00.01 | 1 |
--------------------------------------------------------------------------------------------------------
14 rows selected.
我们看到这个执行计划没有走balanced列上的索引,而是走了和date_je的联合索引。这个可以查看另一篇文档:Note:67522.1 Why is my index not used?
小结一下:
我这里只是简单的从执行计划上看count(*)、count(1)和count(主键)其实是一致,MOS的文档中详细的讲解了Oracle是如何评估执行计划的,也可以使用10053 event查看CBO优化器是如何做出选择的。由于我的功力还不够,对于10053事件还不是很明白,暂时就先不做演示了,要不哪说错了就不好了,这也可以做为以后博客分享的内容。
从这个事情上来看,我们对于一件事情应该做一个深入的研究,有充足的证据来证明,尤其是想要在某一方面有深入发展的时候。