之前生产环境有一部分sql执行出现较大的性能问题，都是由于应该走hash join的走的是filter，而这个问题困扰了小鱼很大一阵子，小鱼是想在sql优化上深入学习的，而为什么子查询没有展开选择走的filter而不是性能更好的hash join了，下面我们先测试，然后再来
之前生产环境有一部分sql执行出现较大的性能问题，都是由于应该走hash join的走的是filter，而这个问题困扰了小鱼很大一阵子，小鱼是想在sql优化上深入学习的，而为什么子查询没有展开选择走的filter而不是性能更好的hash join了，下面我们先测试，然后再来看下生产环境的案例：
数据库版本是10.2.0.4,建立三个表table01、table02、table03
C:\Users\Administrator>sqlplus / as sysdba
SQL*Plus: Release 10.2.0.4.0 - Production on Thu May 15 14:53:07 2014
Copyright (c) 1982, 2007, Oracle. All Rights Reserved.
Connected to:
Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
SQL> create table table01 as select * from dba_objects;
SQL> create table table02 as select * from dba_objects;
SQL> create table table03 as select * from dba_objects;
SQL> select * from table02 b where (object_id in (select object_id from table01
a) or object_id in (select object_id from table03));
50038 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 64347382
-------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 4712 | 814K| 154 (2)| 00:00:02 |
|* 1 | FILTER | | | | | |
| 2 | TABLE ACCESS FULL | TABLE02 | 48330 | 8353K| 154 (2)| 00:00:02 |
|* 3 | TABLE ACCESS FULL | TABLE01 | 512 | 6656 | 2 (0)| 00:00:01 |
|* 4 | TABLE ACCESS FULL| TABLE03 | 440 | 5720 | 3 (0)| 00:00:01 |
-------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter( EXISTS (SELECT /*+ */ 0 FROM "TABLE01" "A" WHERE
"OBJECT_ID"=:B1) OR EXISTS (SELECT /*+ */ 0 FROM "TABLE03" "TABLE
03"
WHERE "OBJECT_ID"=:B2))
3 - filter("OBJECT_ID"=:B1)
4 - filter("OBJECT_ID"=:B1)
Note
-----
- dynamic sampling used for this statement
Statistics
----------------------------------------------------------
65 recursive calls
0 db block gets
17166265 consistent gets
687 physical reads
0 redo size
2565005 bytes sent via SQL*Net to client
37177 bytes received via SQL*Net from client
3337 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
50038 rows processed
这里看出由于子查询中加了or，cbo只能选择走低性能的filter，而其实这个sql我们改写成如下的形式，此时cbo就能展开这个子查询为表连接
SQL> SELECT *
2 FROM table02 b
3 WHERE object_id IN ( (SELECT object_id
4 FROM table01 a)
5 UNION ALL
6 (SELECT object_id FROM table03));
50038 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 3306159213
--------------------------------------------------------------------------------
----------
| Id | Operation | Name | Rows | Bytes |TempSpc| Cost (%CPU)|
Time |
--------------------------------------------------------------------------------
----------
| 0 | SELECT STATEMENT | | 95255 | 17M| | 1008 (1)|
00:00:13 |
|* 1 | HASH JOIN | | 95255 | 17M| 2328K| 1008 (1)|
00:00:13 |
| 2 | VIEW | VW_NSO_1 | 95253 | 1209K| | 306 (1)|
00:00:04 |
| 3 | HASH UNIQUE | | 95253 | 1209K| | 306 (51)|
00:00:04 |
| 4 | UNION-ALL | | | | | |
|
| 5 | TABLE ACCESS FULL| TABLE01 | 51209 | 650K| | 153 (1)|
00:00:02 |
| 6 | TABLE ACCESS FULL| TABLE03 | 44044 | 559K| | 153 (1)|
00:00:02 |
| 7 | TABLE ACCESS FULL | TABLE02 | 48330 | 8353K| | 154 (2)|
00:00:02 |
--------------------------------------------------------------------------------
----------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - access("OBJECT_ID"="$nso_col_1")
Note
-----
- dynamic sampling used for this statement
Statistics
----------------------------------------------------------
0 recursive calls
0 db block gets
5373 consistent gets
0 physical reads
0 redo size
2565005 bytes sent via SQL*Net to client
37177 bytes received via SQL*Net from client
3337 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
50038 rows processed
我们来看看11.2.0.3下是否也需要改写了，小鱼也没试过，测试下吧：
SQL> select * from table02 b where (object_id in (select object_id from table01
2 a) or object_id in (select object_id from table03));
87289 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 1591019701
------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes |TempSpc| Cost (%CPU)| Time |
------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 183K| 38M| | 2092 (1)| 00:00:26 |
|* 1 | HASH JOIN | | 183K| 38M| 4472K| 2092 (1)| 00:00:26 |
| 2 | VIEW | VW_NSO_1 | 183K| 2325K| | 677 (1)| 00:00:09 |
| 3 | HASH UNIQUE | | 183K| 2325K| | 677 (51)| 00:00:09 |
| 4 | UNION-ALL | | | | | | |
| 5 | TABLE ACCESS FULL| TABLE03 | 89910 | 1141K| | 338 (1)| 00:00:05 |
| 6 | TABLE ACCESS FULL| TABLE01 | 93256 | 1183K| | 338 (1)| 00:00:05 |
| 7 | TABLE ACCESS FULL | TABLE02 | 82647 | 16M| | 339 (1)| 00:00:05 |
------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - access("OBJECT_ID"="OBJECT_ID")
Note
-----
- dynamic sampling used for this statement (level=2)
Statistics
----------------------------------------------------------
0 recursive calls
0 db block gets
9445 consistent gets
3720 physical reads
0 redo size
4707818 bytes sent via SQL*Net to client
64529 bytes received via SQL*Net from client
5821 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
87289 rows processed
11.2.0.3版本的cbo优化器居然自动把我们这个or查询组合为一个union all然后hash unique去重为一个view视图然后去和主表做hash join，优化器确是在不断的进步。
类似的子查询无法展开的问题我们还可以再看下面几个例子
Oracle 10.2.0.4版本：
SQL> select * from table02 b where object_id in (select ob
ject_id from table01 a where b.object_id>5000);
45120 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 3332700264
-------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 32 | 5664 | 3182 (1)| 00:00:39 |
|* 1 | FILTER | | | | | |
| 2 | TABLE ACCESS FULL | TABLE02 | 48330 | 8353K| 154 (2)| 00:00:02 |
|* 3 | FILTER | | | | | |
|* 4 | TABLE ACCESS FULL| TABLE01 | 512 | 6656 | 2 (0)| 00:00:01 |
-------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter( EXISTS (SELECT /*+ */ 0 FROM "TABLE01" "A" WHERE
:B1>5000 AND "OBJECT_ID"=:B2))
3 - filter(:B1>5000)
4 - filter("OBJECT_ID"=:B1)
Note
-----
- dynamic sampling used for this statement
Statistics
----------------------------------------------------------
13 recursive calls
0 db block gets
17001967 consistent gets
0 physical reads
0 redo size
2335963 bytes sent via SQL*Net to client
33569 bytes received via SQL*Net from client
3009 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
45120 rows processed
其实这里很可能是个笔误引起的，应该是a.object_id>5000,可能开发人员写成了b.object_id>5000,悲剧的走了filter。
Oracle 11.2.0.3版本：
SQL> SELECT *
2 FROM table02 b
3 WHERE object_id IN (SELECT object_id
4 FROM table01 a
5 WHERE b.object_id > 5000);
82290 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 1361234999
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 4132 | 887K| 678 (1)| 00:00:09 |
|* 1 | HASH JOIN SEMI | | 4132 | 887K| 678 (1)| 00:00:09 |
|* 2 | TABLE ACCESS FULL | TABLE02 | 4132 | 835K| 338 (1)| 00:00:05 |
| 3 | VIEW | VW_NSO_1 | 91760 | 1164K| 339 (1)| 00:00:05 |
|* 4 | TABLE ACCESS FULL| TABLE01 | 91760 | 1164K| 339 (1)| 00:00:05 |
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - access("OBJECT_ID"="OBJECT_ID")
2 - filter("B"."OBJECT_ID">5000)
4 - filter("OBJECT_ID">5000)
Note
-----
- dynamic sampling used for this statement (level=2)
Statistics
----------------------------------------------------------
32 recursive calls
0 db block gets
8579 consistent gets
2480 physical reads
0 redo size
4466004 bytes sent via SQL*Net to client
60855 bytes received via SQL*Net from client
5487 SQL*Net roundtrips to/from client
3 sorts (memory)
0 sorts (disk)
82290 rows processed
可能有朋友提到这里的查询用的是关联列，如果我们用非关联过滤了，其实结果也是一样的
SQL> SELECT *
2 FROM table02 b
3 WHERE object_id IN (SELECT object_id
4 FROM table01 a
5 WHERE b.data_object_id > 5000);
19927 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 1361234999
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 4132 | 887K| 678 (1)| 00:00:09 |
|* 1 | HASH JOIN SEMI | | 4132 | 887K| 678 (1)| 00:00:09 |
|* 2 | TABLE ACCESS FULL | TABLE02 | 4132 | 835K| 339 (1)| 00:00:05 |
| 3 | VIEW | VW_NSO_1 | 93256 | 1183K| 338 (1)| 00:00:05 |
| 4 | TABLE ACCESS FULL| TABLE01 | 93256 | 1183K| 338 (1)| 00:00:05 |
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - access("OBJECT_ID"="OBJECT_ID")
2 - filter("B"."DATA_OBJECT_ID">5000)
Note
-----
- dynamic sampling used for this statement (level=2)
Statistics
----------------------------------------------------------
31 recursive calls
0 db block gets
4491 consistent gets
2480 physical reads
0 redo size
1270866 bytes sent via SQL*Net to client
15128 bytes received via SQL*Net from client
1330 SQL*Net roundtrips to/from client
3 sorts (memory)
0 sorts (disk)
19927 rows processed
这个笔误是常有的事情，或者说开发人员往往写出来的代码只要满足业务逻辑需求，比如小鱼前不久碰见的一个案例也是开发人员写的代码，业务逻辑是满足的，具体代码类似这种查询：
SELECT *
FROM table02 b
WHERE object_id IN (SELECT object_id
FROM table02 a
WHERE b.object_id > 5000);
Execution Plan
----------------------------------------------------------
Plan hash value: 3539933929
-------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 32 | 5664 | 3182 (1)| 00:00:39 |
|* 1 | FILTER | | | | | |
| 2 | TABLE ACCESS FULL | TABLE02 | 48330 | 8353K| 154 (2)| 00:00:02 |
|* 3 | FILTER | | | | | |
|* 4 | TABLE ACCESS FULL| TABLE02 | 483 | 6279 | 2 (0)| 00:00:01 |
-------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter( EXISTS (SELECT /*+ */ 0 FROM "TABLE02" "A" WHERE
:B1>5000 AND "OBJECT_ID"=:B2))
3 - filter(:B1>5000)
4 - filter("OBJECT_ID"=:B1)
Note
-----
- dynamic sampling used for this statement
Statistics
----------------------------------------------------------
14 recursive calls
0 db block gets
17001959 consistent gets
0 physical reads
0 redo size
2336165 bytes sent via SQL*Net to client
33580 bytes received via SQL*Net from client
3010 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
45121 rows processed
这里由于table02 a和table02 b是同一个表，所以这里业务逻辑完全符合，刚开始数据量小，
他们就一直这么写着用着，慢慢得数量大了，由于是10.2.0.5数据库，后面走filter效率极低，极大的拉低了数据库性能，导致系统宕机，而同样的情况在oracle 11g下优化器却改善了很多。
再来看看下面这个查询，有时候我们需要连接两个表，然后分组求出满足某个列聚合函数的数据
Oracle 10.2.0.4版本：
SQL> select * from table02 b where created
here a.object_id=b.object_id group by a.object_id);
no rows selected
Execution Plan
----------------------------------------------------------
Plan hash value: 1461399523
--------------------------------------------------------------------------------
-
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time
|
--------------------------------------------------------------------------------
-
| 0 | SELECT STATEMENT | | 2 | 354 | 3185 (1)| 00:00:39
|
|* 1 | FILTER | | | | |
|
| 2 | TABLE ACCESS FULL | TABLE02 | 48330 | 8353K| 154 (2)| 00:00:02
|
| 3 | SORT GROUP BY NOSORT| | 512 | 11264 | 2 (0)| 00:00:01
|
|* 4 | TABLE ACCESS FULL | TABLE01 | 512 | 11264 | 2 (0)| 00:00:01
|
--------------------------------------------------------------------------------
-
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("CREATED"
"A" WHERE "A"."OBJECT_ID"=:B1 GROUP BY "A"."OBJECT_ID"))
4 - filter("A"."OBJECT_ID"=:B1)
Note
-----
- dynamic sampling used for this statement
Statistics
----------------------------------------------------------
37 recursive calls
0 db block gets
34627266 consistent gets
0 physical reads
0 redo size
1139 bytes sent via SQL*Net to client
481 bytes received via SQL*Net from client
1 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
0 rows processed
而其实这个sql等价于下面这个sql语句，而走的执行计划是更优秀的hash join
SQL> SELECT *
2 FROM table02 b
3 WHERE created
4 FROM table01 a
5 WHERE a.object_id = b.object_id);
no rows selected
Execution Plan
----------------------------------------------------------
Plan hash value: 1435252230
--------------------------------------------------------------------------------
--------
| Id | Operation | Name | Rows | Bytes |TempSpc| Cost (%CPU)| Ti
me |
--------------------------------------------------------------------------------
--------
| 0 | SELECT STATEMENT | | 2560 | 497K| | 827 (1)| 00
:00:10 |
|* 1 | HASH JOIN | | 2560 | 497K| 1704K| 827 (1)| 00
:00:10 |
| 2 | VIEW | VW_SQ_1 | 51209 | 1100K| | 157 (4)| 00
:00:02 |
| 3 | HASH GROUP BY | | 51209 | 1750K| | 157 (4)| 00
:00:02 |
| 4 | TABLE ACCESS FULL| TABLE01 | 51209 | 1750K| | 153 (1)| 00
:00:02 |
| 5 | TABLE ACCESS FULL | TABLE02 | 48330 | 8353K| | 154 (2)| 00
:00:02 |
--------------------------------------------------------------------------------
--------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - access("OBJECT_ID"="B"."OBJECT_ID")
filter("CREATED"
Note
-----
- dynamic sampling used for this statement
Statistics
----------------------------------------------------------
39 recursive calls
0 db block gets
2212 consistent gets
0 physical reads
0 redo size
1139 bytes sent via SQL*Net to client
481 bytes received via SQL*Net from client
1 SQL*Net roundtrips to/from client
7 sorts (memory)
0 sorts (disk)
0 rows processed
oracle 11.2.0.3版本跟10.2.0.4版本也是走的糟糕的filter：
SQL> explain plan for SELECT *
2 FROM table02 b
3 WHERE created
4 FROM table01 a
5 WHERE a.object_id = b.object_id
6 GROUP BY a.object_id);
Explained.
SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------
Plan hash value: 1461399523
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 2 | 414 | 5517 (1)| 00:01:07 |
|* 1 | FILTER | | | | | |
| 2 | TABLE ACCESS FULL | TABLE02 | 82647 | 16M| 339 (1)| 00:00:05 |
| 3 | SORT GROUP BY NOSORT| | 933 | 20526 | 2 (0)| 00:00:01 |
|* 4 | TABLE ACCESS FULL | TABLE01 | 933 | 20526 | 2 (0)| 00:00:01 |
---------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("CREATED"
"A"."OBJECT_ID"=:B1 GROUP BY "A"."OBJECT_ID"))
4 - filter("A"."OBJECT_ID"=:B1)
Note
-----
- dynamic sampling used for this statement (level=2)
22 rows selected.
子查询中包含a.col like b.col的查询
Oracle 10.2.0.4版本：
SQL> SELECT /*+use_hash(a b)*/ *
2 FROM table02 b
3 WHERE object_id IN (SELECT object_id
4 FROM table02 a
5 WHERE a.object_type like b.object_type);
50038 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 337184939
------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 177 | 645K (1)| 02:09:02 |
|* 1 | FILTER | | | | | |
| 2 | TABLE ACCESS FULL| TABLE02 | 48330 | 8353K| 154 (2)| 00:00:02 |
|* 3 | TABLE ACCESS FULL| TABLE02 | 24 | 576 | 14 (0)| 00:00:01 |
------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter( EXISTS (SELECT /*+ */ 0 FROM "TABLE02" "A" WHERE
"OBJECT_ID"=:B1 AND "A"."OBJECT_TYPE" LIKE :B2))
3 - filter("OBJECT_ID"=:B1 AND "A"."OBJECT_TYPE" LIKE :B2)
Note
-----
- dynamic sampling used for this statement
Statistics
----------------------------------------------------------
13 recursive calls
0 db block gets
17165439 consistent gets
0 physical reads
0 redo size
2565005 bytes sent via SQL*Net to client
37177 bytes received via SQL*Net from client
3337 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
50038 rows processed
Oracle 11.2.0.3版本
SQL> explain plan for SELECT *
2 FROM table02 b
3 WHERE object_id IN (SELECT object_id
4 FROM table02 a
5 WHERE a.object_type like b.object_type);
Explained.
SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------
Plan hash value: 337184939
------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 207 | 1447K (1)| 04:49:25 |
|* 1 | FILTER | | | | | |
| 2 | TABLE ACCESS FULL| TABLE02 | 82647 | 16M| 339 (1)| 00:00:05 |
|* 3 | TABLE ACCESS FULL| TABLE02 | 41 | 984 | 18 (0)| 00:00:01 |
------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter( EXISTS (SELECT 0 FROM "TABLE02" "A" WHERE
"OBJECT_ID"=:B1 AND "A"."OBJECT_TYPE" LIKE :B2))
3 - filter("OBJECT_ID"=:B1 AND "A"."OBJECT_TYPE" LIKE :B2)
Note
-----
- dynamic sampling used for this statement (level=2)
21 rows selected.
这里由于有a.object_type like b.object_type，子查询没有展开，那么是否有办法进行优化了，由于hash join算法决定只能用于等值比较，所以这种a.col like b.col是无法转化为hash join的，如果系统里面真有这种业务需求，那么只能采取别的办法来进行优化了。
上面小鱼简单的列出了子查询无法展开的几个现象和现在比较流行的oracle 10gR2和11gR2版本对于子查询的几个变化。下面小鱼用简短的文章来描述下前些天遇见的几个案例：
1 案例开发人员笔误引起的sql性能问题 http://www.dbaxiaoyu.com/archives/1995
小鱼当初没有过分的去分析为什么最后会走filter，而不走hash join，现在通过上面的几个案例测试，想问题原因基本已经知道了，开发人员有个笔误（其实这个sql也是满足业务需求的，对于开发人员可能不叫笔误）
Select * from abstractti0_
Where abstractti0_.TICKET_ID in （（
Select TICKET_ID from tab02 where。。。）
UNION
(SELECT abstractti8_.TICKET_ID
FROM HF_LT_WORKFLOW_ABSTRACTTICKET abstractti8_
WHERE (abstractti0_.CURRENT_STAGE IN
('xqywtj_deal3', 'xqybcgj_deal3'))
AND (abstractti0_.TICKET_ID IN
(SELECT DISTINCT workitemco0_.ENTITY_ID
FROM HF_LTWORKFLOW_ITEM_COMPLETED workitemco0_
WHERE (workitemco0_.BELONG_TO_ID =
'U00000000805')
AND (workitemco0_.WORKFLOWID =
'xqybcgjWorkflow'))))
由于sql较长，接近100多行，这里为了展示方便，就做了一些改写，基本含义就上面这种，子查询中的又出现了主查询的表的过滤条件，而这个数据库版本正是10.2.0.5版本，这个将导致优化器无法展开子查询为表连接，而不得不选择性能较差的filter，最后改掉abstractti0_为abstractti8_，sql基本在几秒内完成，具体大家可以自己看着上面的测试来进行模拟。
2 or 查询导致子查询无法展开 http://www.dbaxiaoyu.com/archives/2048 选择了错误的filter
AND((abstractti0_.TICKET_ID IN
(SELECT abstractti4_.TICKET_ID
FROM HF_LT_WORKFLOW_ABSTRACTTICKET abstractti4_
WHERE (1 =1 )
AND(abstractti4_.TICKET_ID IN
(SELECT DISTINCT workitemco0_.ENTITY_ID
FROM HF_LTWORKFLOW_ITEM_COMPLETED workitemco0_
WHERE (workitemco0_.BELONG_TO_ID='admin' )
AND(workitemco0_.WORKFLOWID ='x qybcgjWorkflow' )
))
AND(abstractti4_.CURRENT_STAGE='ticket_close' )
))
OR(abstractti0_.TICKET_ID IN
(SELECT DISTINCT abstractti6_.TICKET_ID
FROM HF_LTWORKFLOW_ITEM workitem5_,
HF_LT_WORKFLOW_ABSTRACTTICKET abstractti6_,
HF_DM_SAMPLE dmsample7_
WHERE (1 =1 )
AND(abstractti6_.TICKET_ID IN
(SELECT DISTINCT workitemco0_.ENTITY_ID
FROM HF_LTWORKFLOW_ITEM_COMPLETED workitemco0_
WHERE (workitemco0_.BELONG_TO_ID='admin' )
AND(workitemco0_.WORKFLOWID ='xqybcgjWorkflow' )
))
同样由于sql较长，这些报表的sql都是这么长，数据库版本是10.2.0.5版本，由于or的出现，导致了最后也走了filter，对于这类sql的改写小鱼在上面的测试中也提到了改写为union all即可。
关于sql优化需要大量的经验和对cbo的深入理解，而我们当初学习数据库时就知道一个信息，数据库的性能问题80%来源于不合理的sql，希望小鱼自己在sql优化这个路上有一定的建树
Good luck！
原文地址：子查询无法展开的模拟和几个案例, 感谢原作者分享。
如何查证JBL耳机的真假信息Dec 29, 2023 pm 10:54 PMjbl耳机是很多听音乐用户的首选，好评如潮，但是对于假货大家还是非常的害怕的，那么jbl耳机怎么查询真伪来避免这个问题呢？下面就看看怎么查询吧。jbl耳机怎么查询真伪：1、首先进入“中国商品信息验证中心”。2、然后输入查询码，即可查看是否正确从而判断是不是真货。3、也可以去通过耳机声音的清晰度来进行分辨。正牌的耳机声音是非常的清晰的，音质也不会出现改变。假牌的耳机声音会有很多的掺杂，音质也是特别的差。4、大家可以将自己的耳机声音开到最大，看看是不是很和谐，真耳机声音都是一样的。但是假耳机的声音开
买过比特币怎么查询,怎么看自己买没买过比特币Feb 04, 2024 am 08:09 AM现在怎么买比特币的800字公众号文章随着比特币的价格飙升，越来越多的人开始关注比特币，也有越来越多的人想要投资比特币。但是，投资比特币也是有风险的，所以在投资之前，需要了解相关知识，并做好风险控制。那么现在怎么买比特币呢？首先，为了投资比特币，您需要准备一个比特币钱包。比特币钱包是一种软件，用于存储、发送和接收比特币资金。根据您的需求，有多种类型的比特币钱包可供选择，例如硬件钱包、软件钱包和网络钱包。每种钱包都有其独特的特点，您可以根据自己的需求选择适合自己的。其次，你需要选择一个可靠的比特币交
steam怎么查询账号idFeb 07, 2024 pm 12:09 PMSteam是一个整合游戏下载平台，玩家可以在该平台上购买正版的游戏，也可以跟其他玩家进行沟通讨论交流。有小伙伴知道steam怎么查询账号id吗，下面小编就给大家详细介绍一下steam查询账号id的方法，有需要的小伙伴可以来看一看。查询方法：1、双击打开软件，点击上方菜单栏中的"查看"。2、然后点击下方选项列表中的"设置"。3、在打开的窗口界面中，点击左侧栏中的"账户"选项，然后找到右侧中的"账户名称"即可查询到账号id。
PHP如何查询Oracle数据库中的数据Jul 13, 2023 pm 07:34 PMPHP如何查询Oracle数据库中的数据随着互联网时代的到来，网站和应用程序的开发越来越普遍。而数据库作为数据存储和管理的关键技术，也成为了开发者们必备的工具之一。其中，Oracle数据库作为一款功能强大、稳定可靠的关系型数据库管理系统，在企业级应用中得到了广泛应用。而在开发网站或应用程序时，如何使用PHP进行Oracle数据库的查询是一个非常重要的问题。在
PHP表单处理：表单数据查询与筛选Aug 07, 2023 pm 06:17 PMPHP表单处理：表单数据查询与筛选引言在Web开发中，表单是一种重要的交互方式，用户可以通过表单向服务器提交数据并进行进一步的处理。本文将介绍如何使用PHP处理表单数据的查询与筛选功能。表单的设计与提交首先，我们需要设计一个包含查询与筛选功能的表单。常见的表单元素包括输入框、下拉列表、单选框、复选框等，根据具体需求进行设计。用户在提交表单时，会将数据以POS
查询从节点X开始，距离最多为D的子树中的最小权重Aug 25, 2023 am 11:25 AM在进行计算机编程时，有时需要求出源自特定节点的子树的最小权重，条件是该子树不能包含距离指定节点超过D个单位的节点。这个问题出现在各个领域和应用中，包括图论、基于树的算法和网络优化。子树是较大树结构的子集，指定的节点作为子树的根节点。子树包含根节点的所有后代及其连接边。节点的权重是指分配给该节点的特定值，可以表示其重要性、重要性或其他相关指标。在这个问题中，目标是找到子树中所有节点中的最小权重，同时将子树限制在距离根节点最多D个单位的节点。在下面的文章中，我们将深入研究从子树中挖掘最小权重的复杂性
MySQL中如何进行数据的时序存储和查询？Jul 30, 2023 pm 03:09 PMMySQL中如何进行数据的时序存储和查询？在许多应用场景中，对于数据的时序性要求是非常常见的，比如传感器数据的监测、日志记录等。MySQL作为一种常用的关系型数据库，也提供了一些方法来进行时序数据的存储和查询。一种常见的方法是使用时间戳字段来存储数据的时间信息。在MySQL中，可以使用TIMESTAMP或DATETIME类型的字段来存储时间。TIMESTAM
高德地图API文档解读：Java代码实现公交车在线运行状态查询Jul 29, 2023 pm 10:45 PM高德地图API文档解读：Java代码实现公交车在线运行状态查询导语：随着城市的发展，公共交通的重要性越来越凸显出来。人们对公交车的运行状态有着强烈的需求，例如实时到站时间、拥挤程度等信息。高德地图提供了强大的API以满足这方面的需求。本文将解读高德地图API文档，使用Java代码实现公交车在线运行状态查询，并提供代码示例。API概述高德地图API提供了丰富的
Hot AI Tools
Undresser.AI Undress
AI-powered app for creating realistic nude photos
AI Clothes Remover
Online AI tool for removing clothes from photos.
Undress AI Tool
Undress images for free
Clothoff.io
AI clothes remover
AI Hentai Generator
Generate AI Hentai for free.
Hot Article
Hot Tools
Dreamweaver Mac version
Visual web development tools
MinGW - Minimalist GNU for Windows
This project is in the process of being migrated to osdn.net/projects/mingw, you can continue to follow us there. MinGW: A native Windows port of the GNU Compiler Collection (GCC), freely distributable import libraries and header files for building native Windows applications; includes extensions to the MSVC runtime to support C99 functionality. All MinGW software can run on 64-bit Windows platforms.
MantisBT
Mantis is an easy-to-deploy web-based defect tracking tool designed to aid in product defect tracking. It requires PHP, MySQL and a web server. Check out our demo and hosting services.
Atom editor mac version download
The most popular open source editor
Notepad++7.3.1
Easy-to-use and free code editor
