分享程序员开发的那些事...
更新时间：2022-09-13 11:03:20
[20150706]11G谓词推入问题.txt
--生产系统遇到一个sql语句的问题.
--生产系统的sql语句比较复杂,做一个简化的例子来说明问题.来说明自己优化遇到的困惑。
1.建立测试环境:
SCOTT@test> @ver1
PORT_STRING VERSION BANNER
------------------------------ -------------- --------------------------------------------------------------------------------
x86_64/Linux 2.4.xx 11.2.0.3.0 Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
create table t1 as select rownum id1, trunc(rownum/10)+1 id2,cast('t1test' as varchar2(20)) pad1 from xmltable('1 to 100000');
create table t2 as select lpad(rownum,10,'0') id1, cast('t2test' as varchar2(20)) pad2 from xmltable('1 to 100000');
create table t3 as select rownum id1, trunc(dbms_random.value(0,100)) x1,trunc(dbms_random.value(0,200)) x2,cast('t3test' as varchar2(20)) pad3 from xmltable('1 to 100000');
create table t4 as select * from t3 where 1=2;
--t4是空表,主要目的这样与生产系统信息符合.没有什么其它意思.
--另外T2表字段id1是字符型的，问题主要在这里。
create unique index pk_t1 on t1 (ID1);
alter table t1 add constraint pk_t1 primary key (id1);
create index i_t1_id2 on t1 (ID2);
create index pk_t2 on t2 (to_number(ID1));
create index i_t3_id1 on t3 (ID1);
create index i_t4_id1 on t4 (ID1);
--分析表忽略.Method_Opt => 'FOR ALL COLUMNS SIZE 1 '
create view v_t3t4 as select * from t3 union all select * from t4;
create or replace view v_tall as
select t1.id1,t1.id2,t1.pad1,t2.pad2,sum_x1,sum_x2
from t1,t2,
(select id1,sum(x1) sum_x1,sum(x2) sum_x2 from v_t3t4 group by id1) vx
where t1.id1=to_number(t2.id1) and t1.id1=vx.id1;
2.因为我的测试环境遇到cardinality feedback问题：
--参考http://blog.itpub.net/267265/viewspace-1724262/
--我在session关闭cardinality feedback。
SCOTT@test> alter session set "_optimizer_use_feedback"=false ;
Session altered.
SCOTT@test> select * from v_tall where id2=42;
SCOTT@test> @dpc '' outline
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID 09qkq6x7ju4x2, child number 0
-------------------------------------
select * from v_tall where id2=42
Plan hash value: 431511531
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers | OMem | 1Mem | Used-Mem |
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 53 (100)| | 10 |00:00:00.01 | 37 | | | |
| 1 | HASH GROUP BY | | 1 | 10 | 700 | 53 (2)| 00:00:01 | 10 |00:00:00.01 | 37 | 766K| 766K| 1218K (0)|
| 2 | NESTED LOOPS | | 1 | 10 | 700 | 52 (0)| 00:00:01 | 10 |00:00:00.01 | 37 | | | |
| 3 | NESTED LOOPS | | 1 | 10 | 390 | 22 (0)| 00:00:01 | 10 |00:00:00.01 | 12 | | | |
| 4 | TABLE ACCESS BY INDEX ROWID | T1 | 1 | 10 | 160 | 2 (0)| 00:00:01 | 10 |00:00:00.01 | 3 | | | |
|* 5 | INDEX RANGE SCAN | I_T1_ID2 | 1 | 10 | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 2 | | | |
| 6 | TABLE ACCESS BY INDEX ROWID | T2 | 10 | 1 | 23 | 2 (0)| 00:00:01 | 10 |00:00:00.01 | 9 | | | |
|* 7 | INDEX RANGE SCAN | PK_T2 | 10 | 1 | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 8 | | | |
| 8 | VIEW | V_T3T4 | 10 | 1 | 31 | 3 (0)| 00:00:01 | 10 |00:00:00.01 | 25 | | | |
| 9 | UNION ALL PUSHED PREDICATE | | 10 | | | | | 10 |00:00:00.01 | 25 | | | |
| 10 | TABLE ACCESS BY INDEX ROWID| T3 | 10 | 1 | 12 | 2 (0)| 00:00:01 | 10 |00:00:00.01 | 25 | | | |
|* 11 | INDEX RANGE SCAN | I_T3_ID1 | 10 | 1 | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 15 | | | |
| 12 | TABLE ACCESS BY INDEX ROWID| T4 | 10 | 1 | 39 | 1 (0)| 00:00:01 | 0 |00:00:00.01 | 0 | | | |
|* 13 | INDEX RANGE SCAN | I_T4_ID1 | 0 | 1 | | 1 (0)| 00:00:01 | 0 |00:00:00.01 | 0 | | | |
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$25E2CB83
4 - SEL$25E2CB83 / T1@SEL$2
5 - SEL$25E2CB83 / T1@SEL$2
6 - SEL$25E2CB83 / T2@SEL$2
7 - SEL$25E2CB83 / T2@SEL$2
8 - SET$5715CE2E / V_T3T4@SEL$3
9 - SET$5715CE2E
10 - SEL$8E13D68A / T3@SEL$4
11 - SEL$8E13D68A / T3@SEL$4
12 - SEL$9384AC1D / T4@SEL$5
13 - SEL$9384AC1D / T4@SEL$5
Outline Data
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('11.2.0.3')
DB_VERSION('11.2.0.3')
OPT_PARAM('_optimizer_use_feedback' 'false')
ALL_ROWS
OUTLINE_LEAF(@"SEL$8E13D68A")
OUTLINE_LEAF(@"SEL$9384AC1D")
OUTLINE_LEAF(@"SET$5715CE2E")
PUSH_PRED(@"SEL$25E2CB83" "V_T3T4"@"SEL$3" 3)
OUTLINE_LEAF(@"SEL$25E2CB83")
MERGE(@"SEL$3")
OUTLINE(@"SEL$4")
OUTLINE(@"SEL$5")
OUTLINE(@"SET$1")
OUTLINE(@"SEL$25E2CB83")
MERGE(@"SEL$3")
OUTLINE(@"SEL$F5BB74E1")
MERGE(@"SEL$2")
OUTLINE(@"SEL$3")
OUTLINE(@"SEL$1")
OUTLINE(@"SEL$2")
INDEX_RS_ASC(@"SEL$25E2CB83" "T1"@"SEL$2" ("T1"."ID2"))
INDEX_RS_ASC(@"SEL$25E2CB83" "T2"@"SEL$2" "PK_T2")
NO_ACCESS(@"SEL$25E2CB83" "V_T3T4"@"SEL$3")
LEADING(@"SEL$25E2CB83" "T1"@"SEL$2" "T2"@"SEL$2" "V_T3T4"@"SEL$3")
USE_NL(@"SEL$25E2CB83" "T2"@"SEL$2")
USE_NL(@"SEL$25E2CB83" "V_T3T4"@"SEL$3")
USE_HASH_AGGREGATION(@"SEL$25E2CB83")
INDEX_RS_ASC(@"SEL$9384AC1D" "T4"@"SEL$5" ("T4"."ID1"))
INDEX_RS_ASC(@"SEL$8E13D68A" "T3"@"SEL$4" ("T3"."ID1"))
END_OUTLINE_DATA
*/
Predicate Information (identified by operation id):
---------------------------------------------------
5 - access("T1"."ID2"=42)
7 - access("T1"."ID1"="T2"."SYS_NC00003$")
11 - access("ID1"="T1"."ID1")
13 - access("ID1"="T1"."ID1")
--使用谓词推入。 PUSH_PRED(@"SEL$25E2CB83" "V_T3T4"@"SEL$3" 3)
3.但是如果定义成：
create or replace view v_tallx as
select t1.id1,t1.id2,t1.pad1,t2.pad2,sum_x1,sum_x2,(select dname from dept where deptno=t1.id1) dname
from t1,t2,
(select id1,sum(x1) sum_x1,sum(x2) sum_x2 from v_t3t4 group by id1) vx
where t1.id1=to_number(t2.id1) and t1.id1=vx.id1;
--也就是我加入一个标量子查询，问题出现：
select * from v_tallx where id2=2;
SCOTT@test> @dpc '' outline
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID 8j6fp0bc8d9xd, child number 0
-------------------------------------
select * from v_tallx where id2=2
Plan hash value: 1993492796
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes|E-Temp | Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers | OMem | 1Mem | Used-Mem |
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | | 976 (100)| | 10 |00:00:01.47 | 341 | | | |
| 1 | TABLE ACCESS BY INDEX ROWID | DEPT | 10 | 1 | 13 | | 1 (0)| 00:00:01 | 1 |00:00:00.01 | 5 | | | |
|* 2 | INDEX UNIQUE SCAN | PK_DEPT | 10 | 1 | | | 0 (0)| | 1 |00:00:00.01 | 4 | | | |
| 3 | NESTED LOOPS | | 1 | 10 | 590 | | 976 (1)| 00:00:01 | 10 |00:00:01.47 | 341 | | | |
| 4 | MERGE JOIN | | 1 | 10 | 470 | | 956 (1)| 00:00:01 | 10 |00:00:01.47 | 330 | | | |
| 5 | SORT JOIN | | 1 | 100K| 3027K| | 953 (1)| 00:00:01 | 20 |00:00:01.47 | 327 | 3667K| 828K| 3259K (0)|
| 6 | VIEW | | 1 | 100K| 3027K| | 953 (1)| 00:00:01 | 100K|00:00:01.16 | 327 | | | |
| 7 | HASH GROUP BY | | 1 | 100K| 3027K| 4336K| 953 (1)| 00:00:01 | 100K|00:00:00.96 | 327 | 8785K| 2233K| 9825K (0)|
| 8 | VIEW | V_T3T4 | 1 | 100K| 3027K| | 96 (2)| 00:00:01 | 100K|00:00:00.69 | 327 | | | | [0/29437]
| 9 | UNION-ALL | | 1 | | | | | | 100K|00:00:00.49 | 327 | | | |
| 10 | TABLE ACCESS FULL | T3 | 1 | 100K| 1171K| | 94 (2)| 00:00:01 | 100K|00:00:00.10 | 327 | | | |
| 11 | TABLE ACCESS FULL | T4 | 1 | 1 | 39 | | 2 (0)| 00:00:01 | 0 |00:00:00.01 | 0 | | | |
|* 12 | SORT JOIN | | 20 | 10 | 160 | | 3 (34)| 00:00:01 | 10 |00:00:00.01 | 3 | 2048 | 2048 | 2048 (0)|
| 13 | TABLE ACCESS BY INDEX ROWID| T1 | 1 | 10 | 160 | | 2 (0)| 00:00:01 | 10 |00:00:00.01 | 3 | | | |
|* 14 | INDEX RANGE SCAN | I_T1_ID2 | 1 | 10 | | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 2 | | | |
| 15 | TABLE ACCESS BY INDEX ROWID | T2 | 10 | 1 | 12 | | 2 (0)| 00:00:01 | 10 |00:00:00.01 | 11 | | | |
|* 16 | INDEX RANGE SCAN | PK_T2 | 10 | 1 | | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 9 | | | |
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$3 / DEPT@SEL$3
2 - SEL$3 / DEPT@SEL$3
3 - SEL$F5BB74E1
6 - SEL$4 / VX@SEL$2
7 - SEL$4
8 - SET$1 / V_T3T4@SEL$4
9 - SET$1
10 - SEL$5 / T3@SEL$5
11 - SEL$6 / T4@SEL$6
13 - SEL$F5BB74E1 / T1@SEL$2
14 - SEL$F5BB74E1 / T1@SEL$2
15 - SEL$F5BB74E1 / T2@SEL$2
16 - SEL$F5BB74E1 / T2@SEL$2
Outline Data
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('11.2.0.3')
DB_VERSION('11.2.0.3')
OPT_PARAM('_optimizer_use_feedback' 'false')
ALL_ROWS
OUTLINE_LEAF(@"SEL$3")
OUTLINE_LEAF(@"SEL$5")
OUTLINE_LEAF(@"SEL$6")
OUTLINE_LEAF(@"SET$1")
OUTLINE_LEAF(@"SEL$4")
OUTLINE_LEAF(@"SEL$F5BB74E1")
MERGE(@"SEL$2")
OUTLINE(@"SEL$1")
OUTLINE(@"SEL$2")
NO_ACCESS(@"SEL$F5BB74E1" "VX"@"SEL$2")
INDEX_RS_ASC(@"SEL$F5BB74E1" "T1"@"SEL$2" ("T1"."ID2"))
INDEX_RS_ASC(@"SEL$F5BB74E1" "T2"@"SEL$2" "PK_T2")
LEADING(@"SEL$F5BB74E1" "VX"@"SEL$2" "T1"@"SEL$2" "T2"@"SEL$2")
USE_MERGE(@"SEL$F5BB74E1" "T1"@"SEL$2")
USE_NL(@"SEL$F5BB74E1" "T2"@"SEL$2")
NO_ACCESS(@"SEL$4" "V_T3T4"@"SEL$4")
USE_HASH_AGGREGATION(@"SEL$4")
FULL(@"SEL$6" "T4"@"SEL$6")
FULL(@"SEL$5" "T3"@"SEL$5")
INDEX_RS_ASC(@"SEL$3" "DEPT"@"SEL$3" ("DEPT"."DEPTNO"))
END_OUTLINE_DATA
*/
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("DEPTNO"=:B1)
12 - access("T1"."ID1"="VX"."ID1")
filter("T1"."ID1"="VX"."ID1")
14 - access("T1"."ID2"=2)
16 - access("T1"."ID1"="T2"."SYS_NC00003$")
--可以发现T3,T4是全表扫描。谓词推入失效！
--如果写成这样：
select (select dname from dept where deptno=a.id1) ,a.* from v_tall a where id2=2;
SCOTT@test> @dpc '' outline
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID bq98puv7ac22y, child number 1
-------------------------------------
select (select dname from dept where deptno=a.id1) ,a.* from v_tall a
where id2=2
Plan hash value: 1993492796
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes|E-Temp | Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers | OMem | 1Mem | Used-Mem |
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | | 976 (100)| | 10 |00:00:01.48 | 341 | | | |
| 1 | TABLE ACCESS BY INDEX ROWID | DEPT | 10 | 1 | 13 | | 1 (0)| 00:00:01 | 1 |00:00:00.01 | 5 | | | |
|* 2 | INDEX UNIQUE SCAN | PK_DEPT | 10 | 1 | | | 0 (0)| | 1 |00:00:00.01 | 4 | | | |
| 3 | NESTED LOOPS | | 1 | 10 | 590 | | 976 (1)| 00:00:01 | 10 |00:00:01.48 | 341 | | | |
| 4 | MERGE JOIN | | 1 | 10 | 470 | | 956 (1)| 00:00:01 | 10 |00:00:01.48 | 330 | | | |
| 5 | SORT JOIN | | 1 | 100K| 3027K| | 953 (1)| 00:00:01 | 20 |00:00:01.48 | 327 | 3667K| 828K| 3259K (0)|
| 6 | VIEW | | 1 | 100K| 3027K| | 953 (1)| 00:00:01 | 100K|00:00:01.16 | 327 | | | |
| 7 | HASH GROUP BY | | 1 | 100K| 3027K| 4336K| 953 (1)| 00:00:01 | 100K|00:00:00.96 | 327 | 8785K| 2233K| 9809K (0)|
| 8 | VIEW | V_T3T4 | 1 | 100K| 3027K| | 96 (2)| 00:00:01 | 100K|00:00:00.69 | 327 | | | | [0/27892]
| 9 | UNION-ALL | | 1 | | | | | | 100K|00:00:00.49 | 327 | | | |
| 10 | TABLE ACCESS FULL | T3 | 1 | 100K| 1171K| | 94 (2)| 00:00:01 | 100K|00:00:00.10 | 327 | | | |
| 11 | TABLE ACCESS FULL | T4 | 1 | 1 | 39 | | 2 (0)| 00:00:01 | 0 |00:00:00.01 | 0 | | | |
|* 12 | SORT JOIN | | 20 | 10 | 160 | | 3 (34)| 00:00:01 | 10 |00:00:00.01 | 3 | 2048 | 2048 | 2048 (0)|
| 13 | TABLE ACCESS BY INDEX ROWID| T1 | 1 | 10 | 160 | | 2 (0)| 00:00:01 | 10 |00:00:00.01 | 3 | | | |
|* 14 | INDEX RANGE SCAN | I_T1_ID2 | 1 | 10 | | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 2 | | | |
| 15 | TABLE ACCESS BY INDEX ROWID | T2 | 10 | 1 | 12 | | 2 (0)| 00:00:01 | 10 |00:00:00.01 | 11 | | | |
|* 16 | INDEX RANGE SCAN | PK_T2 | 10 | 1 | | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 9 | | | |
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$2 / DEPT@SEL$2
2 - SEL$2 / DEPT@SEL$2
3 - SEL$8976F1A6
6 - SEL$4 / VX@SEL$3
7 - SEL$4
8 - SET$1 / V_T3T4@SEL$4
9 - SET$1
10 - SEL$5 / T3@SEL$5
11 - SEL$6 / T4@SEL$6
13 - SEL$8976F1A6 / T1@SEL$3
14 - SEL$8976F1A6 / T1@SEL$3
15 - SEL$8976F1A6 / T2@SEL$3
16 - SEL$8976F1A6 / T2@SEL$3
Outline Data
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('11.2.0.3')
DB_VERSION('11.2.0.3')
OPT_PARAM('_optimizer_use_feedback' 'false')
ALL_ROWS
OUTLINE_LEAF(@"SEL$2")
OUTLINE_LEAF(@"SEL$5")
OUTLINE_LEAF(@"SEL$6")
OUTLINE_LEAF(@"SET$1")
OUTLINE_LEAF(@"SEL$4")
OUTLINE_LEAF(@"SEL$8976F1A6")
MERGE(@"SEL$3")
OUTLINE(@"SEL$1")
OUTLINE(@"SEL$3")
NO_ACCESS(@"SEL$8976F1A6" "VX"@"SEL$3")
INDEX_RS_ASC(@"SEL$8976F1A6" "T1"@"SEL$3" ("T1"."ID2"))
INDEX_RS_ASC(@"SEL$8976F1A6" "T2"@"SEL$3" "PK_T2")
LEADING(@"SEL$8976F1A6" "VX"@"SEL$3" "T1"@"SEL$3" "T2"@"SEL$3")
USE_MERGE(@"SEL$8976F1A6" "T1"@"SEL$3")
USE_NL(@"SEL$8976F1A6" "T2"@"SEL$3")
NO_ACCESS(@"SEL$4" "V_T3T4"@"SEL$4")
USE_HASH_AGGREGATION(@"SEL$4")
FULL(@"SEL$6" "T4"@"SEL$6")
FULL(@"SEL$5" "T3"@"SEL$5")
INDEX_RS_ASC(@"SEL$2" "DEPT"@"SEL$2" ("DEPT"."DEPTNO"))
END_OUTLINE_DATA
*/
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("DEPTNO"=:B1)
12 - access("T1"."ID1"="VX"."ID1")
filter("T1"."ID1"="VX"."ID1")
14 - access("T1"."ID2"=2)
16 - access("T1"."ID1"="T2"."SYS_NC00003$")
--问题依旧!那位知道如何解决这个问题。