[20190524]使用use_concat or_expand提示優化.txt
[20190524]使用use_concat or_expand提示優化.txt
--//上午看了連結https://connor-mcdonald.com/2019/05/22/being-generous-to-the-optimizer，突然想起我們生產系統類似語句。
--//現在想想覺得開發的想象力太豐富，寫這些語句是否考慮長期執行導致的結果。對方例子相對簡單，我優化的例子簡直就是變態。
--//連結:[20150814]使用use_concat提示.txt => http://blog.itpub.net/267265/viewspace-1771727/
--//實際上看了馬上想到使用use_concat or_expand提示優化sql語句.同時看了連結
--//https://jonathanlewis.wordpress.com/2019/05/22/danger-hints/，一起測試看看。
1.環境：
SCOTT@test01p> @ ver1
PORT_STRING VERSION BANNER CON_ID
------------------------------ -------------- -------------------------------------------------------------------------------- ----------
IBMPC/WIN_NT64-9.1.0 12.2.0.1.0 Oracle Database 12c Enterprise Edition Release 12.2.0.1.0 - 64bit Production 0
create table address ( street int, suburb int, post_code int, data char(100));
insert into address select mod(rownum,1e4), mod(rownum,10), mod(rownum,1e2), rownum from dual connect by level <= 1e5;
commit;
exec dbms_stats.gather_table_stats('','ADDRESS')
create index i_address_stress on address ( street );
create index i_address_suburb on address ( suburb );
create index i_address_post_code on address ( post_code );
2.測試:
variable val number = 6
variable choice number = 1
alter session set statistics_level = all;
SCOTT@test01p> select data from address where ( :choice = 1 and street = :val ) or ( :choice = 2 and suburb = :val );
DATA
-----
6
10006
20006
30006
40006
50006
60006
70006
80006
90006
10 rows selected.
Plan hash value: 3645838471
-----------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
-----------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 445 (100)| | 10 |00:00:00.02 | 1636 |
|* 1 | TABLE ACCESS FULL| ADDRESS | 1 | 100 | 10800 | 445 (1)| 00:00:01 | 10 |00:00:00.02 | 1636 |
-----------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$1 / ADDRESS@SEL$1
Peeked Binds (identified by position):
--------------------------------------
2 - :2 (NUMBER): 6
4 - :2 (NUMBER, Primary=2)
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter(((:CHOICE=2 AND "SUBURB"=:VAL) OR ("STREET"=:VAL AND :CHOICE=1)))
--//選擇全表掃描.加入提示:/*+ or_expand(@sel$1) */
select /*+ or_expand(@sel$1) */ data from address where ( :choice = 1 and street = :val ) or ( :choice = 2 and suburb = :val );
SCOTT@test01p> @ dpc '' outline
Plan hash value: 1427591975
--------------------------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers | Reads |
--------------------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 456 (100)| | 10 |00:00:00.06 | 13 | 4 |
| 1 | VIEW | VW_ORE_B7380F92 | 1 | 10010 | 997K| 456 (1)| 00:00:01 | 10 |00:00:00.06 | 13 | 4 |
| 2 | UNION-ALL | | 1 | | | | | 10 |00:00:00.06 | 13 | 4 |
|* 3 | FILTER | | 1 | | | | | 10 |00:00:00.06 | 13 | 4 |
| 4 | TABLE ACCESS BY INDEX ROWID BATCHED| ADDRESS | 1 | 10 | 1050 | 11 (0)| 00:00:01 | 10 |00:00:00.06 | 13 | 4 |
|* 5 | INDEX RANGE SCAN | I_ADDRESS_STRESS | 1 | 10 | | 1 (0)| 00:00:01 | 10 |00:00:00.06 | 3 | 4 |
|* 6 | FILTER | | 1 | | | | | 0 |00:00:00.01 | 0 | 0 |
|* 7 | TABLE ACCESS FULL | ADDRESS | 0 | 10000 | 1054K| 445 (1)| 00:00:01 | 0 |00:00:00.01 | 0 | 0 |
--------------------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SET$9162BF3C / VW_ORE_B7380F92@SEL$B7380F92
2 - SET$9162BF3C
3 - SET$9162BF3C_1
4 - SET$9162BF3C_1 / ADDRESS@SEL$1
5 - SET$9162BF3C_1 / ADDRESS@SEL$1
6 - SET$9162BF3C_2
7 - SET$9162BF3C_2 / ADDRESS@SEL$1
Outline Data
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('12.2.0.1')
DB_VERSION('12.2.0.1')
ALL_ROWS
OUTLINE_LEAF(@"SET$9162BF3C_2")
OUTLINE_LEAF(@"SET$9162BF3C_1")
OUTLINE_LEAF(@"SET$9162BF3C")
OR_EXPAND(@"SEL$1" (1) (2))
OUTLINE_LEAF(@"SEL$B7380F92")
OUTLINE(@"SET$9162BF3C")
OR_EXPAND(@"SEL$1" (1) (2))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
OUTLINE(@"SEL$1")
NO_ACCESS(@"SEL$B7380F92" "VW_ORE_B7380F92"@"SEL$B7380F92")
INDEX_RS_ASC(@"SET$9162BF3C_1" "ADDRESS"@"SEL$1" ("ADDRESS"."STREET"))
BATCH_TABLE_ACCESS_BY_ROWID(@"SET$9162BF3C_1" "ADDRESS"@"SEL$1")
FULL(@"SET$9162BF3C_2" "ADDRESS"@"SEL$1")
END_OUTLINE_DATA
*/
Peeked Binds (identified by position):
--------------------------------------
2 - :2 (NUMBER): 6
4 - :2 (NUMBER, Primary=2)
Predicate Information (identified by operation id):
---------------------------------------------------
3 - filter(:CHOICE=1)
5 - access("STREET"=:VAL)
6 - filter(:CHOICE=2)
7 - filter(("SUBURB"=:VAL AND (LNNVL(:CHOICE=1) OR LNNVL("STREET"=:VAL))))
--//注意看下劃線.
--//但是使用下劃線提示,改動程式碼的情況下如何呢?
SELECT /*+ or_expand(@sel$1 (1) (2) ) */ data
FROM address
WHERE ( :choice = 1 AND street = :val )
OR ( :choice = 2 AND suburb = :val )
OR ( :choice = 3 AND post_code = :val);
SCOTT@test01p> @ dpc '' outline
...
Plan hash value: 1427591975
-----------------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
-----------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 456 (100)| | 10 |00:00:00.01 | 13 |
| 1 | VIEW | VW_ORE_B7380F92 | 1 | 10010 | 997K| 456 (1)| 00:00:01 | 10 |00:00:00.01 | 13 |
| 2 | UNION-ALL | | 1 | | | | | 10 |00:00:00.01 | 13 |
|* 3 | FILTER | | 1 | | | | | 10 |00:00:00.01 | 13 |
| 4 | TABLE ACCESS BY INDEX ROWID BATCHED| ADDRESS | 1 | 10 | 1050 | 11 (0)| 00:00:01 | 10 |00:00:00.01 | 13 |
|* 5 | INDEX RANGE SCAN | I_ADDRESS_STRESS | 1 | 10 | | 1 (0)| 00:00:01 | 10 |00:00:00.01 | 3 |
|* 6 | FILTER | | 1 | | | | | 0 |00:00:00.01 | 0 |
|* 7 | TABLE ACCESS FULL | ADDRESS | 0 | 10000 | 1054K| 445 (1)| 00:00:01 | 0 |00:00:00.01 | 0 |
-----------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SET$9162BF3C / VW_ORE_B7380F92@SEL$B7380F92
2 - SET$9162BF3C
3 - SET$9162BF3C_1
4 - SET$9162BF3C_1 / ADDRESS@SEL$1
5 - SET$9162BF3C_1 / ADDRESS@SEL$1
6 - SET$9162BF3C_2
7 - SET$9162BF3C_2 / ADDRESS@SEL$1
Outline Data
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('12.2.0.1')
DB_VERSION('12.2.0.1')
ALL_ROWS
OUTLINE_LEAF(@"SET$9162BF3C_2")
OUTLINE_LEAF(@"SET$9162BF3C_1")
OUTLINE_LEAF(@"SET$9162BF3C")
OR_EXPAND(@"SEL$1" (1) (2))
OUTLINE_LEAF(@"SEL$B7380F92")
OUTLINE(@"SET$9162BF3C")
OR_EXPAND(@"SEL$1" (1) (2))
OUTLINE(@"SEL$1")
NO_ACCESS(@"SEL$B7380F92" "VW_ORE_B7380F92"@"SEL$B7380F92")
INDEX_RS_ASC(@"SET$9162BF3C_1" "ADDRESS"@"SEL$1" ("ADDRESS"."STREET"))
BATCH_TABLE_ACCESS_BY_ROWID(@"SET$9162BF3C_1" "ADDRESS"@"SEL$1")
FULL(@"SET$9162BF3C_2" "ADDRESS"@"SEL$1")
END_OUTLINE_DATA
*/
Peeked Binds (identified by position):
--------------------------------------
2 - :2 (NUMBER): 6
4 - :2 (NUMBER, Primary=2)
Predicate Information (identified by operation id):
---------------------------------------------------
3 - filter(:CHOICE=1)
5 - access("STREET"=:VAL)
6 - filter(:CHOICE=2)
7 - filter(("SUBURB"=:VAL AND (LNNVL(:CHOICE=1) OR LNNVL("STREET"=:VAL))))
--//正像連結講的那樣如果增加1個或條件,導致執行計劃變得不合理.實際上跟嚴重的是查詢發生了錯誤.
--//如果仔細看Predicate Information 就很容易發現沒有:CHOICE=1的filter.如果查詢:
SCOTT@test01p> variable choice number = 3
SELECT /*+ or_expand(@sel$1 (1) (2) ) */ data
FROM address
WHERE ( :choice = 1 AND street = :val )
OR ( :choice = 2 AND suburb = :val )
OR ( :choice = 3 AND post_code = :val);
no rows selected.
--//取消提示:
SELECT data FROM address WHERE ( :choice = 1 AND street = :val ) OR ( :choice = 2 AND suburb = :val ) OR ( :choice = 3 AND post_code = :val);
...
1000 rows selected.
--//兩者的結果集不一樣.明顯這個是一個bug.
修改如下:
SELECT /*+ or_expand(@sel$1 (1) (2) (3) ) */ data
FROM address
WHERE ( :choice = 1 AND street = :val )
OR ( :choice = 2 AND suburb = :val )
OR ( :choice = 3 AND post_code = :val);
...
--//注使用提示/*+ or_expand(@sel$1 ) */結果是正確的.
SCOTT@test01p> @ dpc '' outline
Plan hash value: 3525475520
-----------------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
-----------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 900 (100)| | 1000 |00:00:00.02 | 1640 |
| 1 | VIEW | VW_ORE_B7380F92 | 1 | 11009 | 1096K| 900 (1)| 00:00:01 | 1000 |00:00:00.02 | 1640 |
| 2 | UNION-ALL | | 1 | | | | | 1000 |00:00:00.02 | 1640 |
|* 3 | FILTER | | 1 | | | | | 0 |00:00:00.01 | 0 |
| 4 | TABLE ACCESS BY INDEX ROWID BATCHED| ADDRESS | 0 | 10 | 1050 | 11 (0)| 00:00:01 | 0 |00:00:00.01 | 0 |
|* 5 | INDEX RANGE SCAN | I_ADDRESS_STRESS | 0 | 10 | | 1 (0)| 00:00:01 | 0 |00:00:00.01 | 0 |
|* 6 | FILTER | | 1 | | | | | 0 |00:00:00.01 | 0 |
|* 7 | TABLE ACCESS FULL | ADDRESS | 0 | 10000 | 1054K| 445 (1)| 00:00:01 | 0 |00:00:00.01 | 0 |
|* 8 | FILTER | | 1 | | | | | 1000 |00:00:00.02 | 1640 |
|* 9 | TABLE ACCESS FULL | ADDRESS | 1 | 999 | 108K| 445 (1)| 00:00:01 | 1000 |00:00:00.02 | 1640 |
-----------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SET$49E1C21B / VW_ORE_B7380F92@SEL$B7380F92
2 - SET$49E1C21B
3 - SET$49E1C21B_1
4 - SET$49E1C21B_1 / ADDRESS@SEL$1
5 - SET$49E1C21B_1 / ADDRESS@SEL$1
6 - SET$49E1C21B_2
7 - SET$49E1C21B_2 / ADDRESS@SEL$1
8 - SET$49E1C21B_3
9 - SET$49E1C21B_3 / ADDRESS@SEL$1
Outline Data
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('12.2.0.1')
DB_VERSION('12.2.0.1')
ALL_ROWS
OUTLINE_LEAF(@"SET$49E1C21B_3")
OUTLINE_LEAF(@"SET$49E1C21B_2")
OUTLINE_LEAF(@"SET$49E1C21B_1")
OUTLINE_LEAF(@"SET$49E1C21B")
OR_EXPAND(@"SEL$1" (1) (2) (3))
OUTLINE_LEAF(@"SEL$B7380F92")
OUTLINE(@"SET$49E1C21B")
OR_EXPAND(@"SEL$1" (1) (2) (3))
OUTLINE(@"SEL$1")
NO_ACCESS(@"SEL$B7380F92" "VW_ORE_B7380F92"@"SEL$B7380F92")
INDEX_RS_ASC(@"SET$49E1C21B_1" "ADDRESS"@"SEL$1" ("ADDRESS"."STREET"))
BATCH_TABLE_ACCESS_BY_ROWID(@"SET$49E1C21B_1" "ADDRESS"@"SEL$1")
FULL(@"SET$49E1C21B_2" "ADDRESS"@"SEL$1")
FULL(@"SET$49E1C21B_3" "ADDRESS"@"SEL$1")
END_OUTLINE_DATA
*/
Peeked Binds (identified by position):
--------------------------------------
2 - :2 (NUMBER): 6
4 - :2 (NUMBER, Primary=2)
6 - :2 (NUMBER, Primary=2)
Predicate Information (identified by operation id):
---------------------------------------------------
3 - filter(:CHOICE=1)
5 - access("STREET"=:VAL)
6 - filter(:CHOICE=2)
7 - filter(("SUBURB"=:VAL AND (LNNVL(:CHOICE=1) OR LNNVL("STREET"=:VAL))))
8 - filter(:CHOICE=3)
9 - filter(("POST_CODE"=:VAL AND (LNNVL(:CHOICE=1) OR LNNVL("STREET"=:VAL)) AND (LNNVL(:CHOICE=2) OR LNNVL("SUBURB"=:VAL))))
3.測試使用use_concate看看:
set linesize 100
SELECT /*+ use_concat */ data
FROM address
WHERE ( :choice = 1 AND street = :val )
OR ( :choice = 2 AND suburb = :val )
OR ( :choice = 3 AND post_code = :val);
SCOTT@test01p> @ dpc '' outline
...
Plan hash value: 2048882018
----------------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
----------------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 900 (100)| | 1000 |00:00:00.01 | 1640 |
| 1 | CONCATENATION | | 1 | | | | | 1000 |00:00:00.01 | 1640 |
|* 2 | FILTER | | 1 | | | | | 0 |00:00:00.01 | 0 |
| 3 | TABLE ACCESS BY INDEX ROWID BATCHED| ADDRESS | 0 | 10 | 1110 | 11 (0)| 00:00:01 | 0 |00:00:00.01 | 0 |
|* 4 | INDEX RANGE SCAN | I_ADDRESS_STRESS | 0 | 10 | | 1 (0)| 00:00:01 | 0 |00:00:00.01 | 0 |
|* 5 | FILTER | | 1 | | | | | 1000 |00:00:00.01 | 1640 |
|* 6 | TABLE ACCESS FULL | ADDRESS | 1 | 1000 | 108K| 445 (1)| 00:00:01 | 1000 |00:00:00.01 | 1640 |
|* 7 | FILTER | | 1 | | | | | 0 |00:00:00.01 | 0 |
|* 8 | TABLE ACCESS FULL | ADDRESS | 0 | 9999 | 1083K| 445 (1)| 00:00:01 | 0 |00:00:00.01 | 0 |
----------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$1
3 - SEL$1_1 / ADDRESS@SEL$1
4 - SEL$1_1 / ADDRESS@SEL$1
6 - SEL$1_2 / ADDRESS@SEL$1_2
8 - SEL$1_3 / ADDRESS@SEL$1_3
Outline Data
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('12.2.0.1')
DB_VERSION('12.2.0.1')
ALL_ROWS
OUTLINE_LEAF(@"SEL$1")
OUTLINE_LEAF(@"SEL$1_1")
USE_CONCAT(@"SEL$1" 8 OR_PREDICATES(1) PREDICATE_REORDERS((5 2) (6 3) (7 4) (8 5) (9 6) (10 7) (2 8) (4 9) (3 10)))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
--//始終不明白裡面那一串表示什麼?
OUTLINE_LEAF(@"SEL$1_2")
OUTLINE_LEAF(@"SEL$1_3")
OUTLINE(@"SEL$1")
INDEX_RS_ASC(@"SEL$1_1" "ADDRESS"@"SEL$1" ("ADDRESS"."STREET"))
BATCH_TABLE_ACCESS_BY_ROWID(@"SEL$1_1" "ADDRESS"@"SEL$1")
FULL(@"SEL$1_2" "ADDRESS"@"SEL$1_2")
FULL(@"SEL$1_3" "ADDRESS"@"SEL$1_3")
END_OUTLINE_DATA
*/
Peeked Binds (identified by position):
--------------------------------------
2 - :2 (NUMBER): 6
4 - :2 (NUMBER, Primary=2)
6 - :2 (NUMBER, Primary=2)
Predicate Information (identified by operation id):
---------------------------------------------------
2 - filter(:CHOICE=1)
4 - access("STREET"=:VAL)
5 - filter(:CHOICE=3)
6 - filter(("POST_CODE"=:VAL AND (LNNVL("STREET"=:VAL) OR LNNVL(:CHOICE=1))))
7 - filter(:CHOICE=2)
8 - filter(("SUBURB"=:VAL AND (LNNVL(:CHOICE=3) OR LNNVL("POST_CODE"=:VAL)) AND (LNNVL("STREET"=:VAL) OR LNNVL(:CHOICE=1))))
來自 “ ITPUB部落格 ” ，連結：http://blog.itpub.net/267265/viewspace-2645474/，如需轉載，請註明出處，否則將追究法律責任。
相關文章
- [20181220]使用提示OR_EXPAND優化.txt優化
- [20181219]不能使用USE_CONCAT優化例子.txt優化
- [20190524]淺談模糊查詢.txt
- [20190524]sqlplus 與輸出&.txtSQL
- [20170328]使用with優化1例.txt優化
- 優化器提示優化
- 關於優化器提示使用逗號的語法優化
- SEO優化-robots.txt解讀優化
- SQLServer效能優化之查詢提示SQLServer優化
- Bad SQL 優化加提示 /*+ordered */SQL優化
- [20181119]使用sql profile優化問題.txtSQL優化
- 總結SQL語句中的優化提示SQL優化
- [20210408]max優化.txt優化
- [20170601]distinct的優化.txt優化
- [20170104]一條sql優化.txtSQL優化
- 如何針對SEO優化您的WordPress Robots.txt優化
- SQL優化案例-使用with as優化Subquery Unnesting（七）SQL優化
- Oracle優化案例-使用with as優化Subquery Unnesting（七）Oracle優化
- 使用explain優化sqlAI優化SQL
- EntityFramework使用及優化Framework優化
- [20190624]12c group by優化 .txt優化
- [20220428]優化的困惑12.txt優化
- [20210203]max優化的困惑.txt優化
- [20201224]sql優化困惑.txtSQL優化
- [20131204]sql語句優化.txtSQL優化
- [20151221]sql語句優化.txtSQL優化
- Econsultancy：站內搜尋優化的5點提示優化
- 使用優化實用工具來優化SQL Server效能優化SQLServer
- # Kotlin使用優化（四）Kotlin優化
- Hive篇---Hive使用優化Hive優化
- 使用索引優化StopKey索引優化TopK
- Hive使用Calcite CBO優化流程及SQL優化實戰Hive優化SQL
- [20200401]優化的困惑5.txt優化
- [20200408]優化的困惑6.txt優化
- [20220507]優化的困惑13.txt優化
- [20201210]sql語句優化.txtSQL優化
- [20140301]直方圖與優化.txt直方圖優化
- [20150611]優化sql遇到問題.txt優化SQL