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：
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。
免责声明：本站发布的内容（图片、视频和文字）以原创、转载和分享为主，文章观点不代表本网站立场，如果涉及侵权请联系站长邮箱：is@yisu.com进行举报，并提供相关证据，一经查实，将立刻删除涉嫌侵权内容。
