Comments on: connect by的性能分析 http://www.os2ora.com/connect-by-performance-tunning/ 专注于现实世界Oracle数据库的高性能，高可扩展性与新一代数据库Exadata架构 Fri, 20 Aug 2010 08:28:04 -0700 http://wordpress.org/?v=2.8 hourly 1 By: Validba'S Home » 求两个时间点之间的时间序列问题 http://www.os2ora.com/connect-by-performance-tunning/comment-page-1/#comment-678 Validba'S Home » 求两个时间点之间的时间序列问题 Fri, 23 Jul 2010 09:47:11 +0000 http://www.os2ora.com/connect-by-performance-tunning/#comment-678 [...] 最后从http://www.os2ora.com/how-to-return-continuous-dates-between-two-dates/发现一个同样的案例,OS2提供了一个非常不错的方法：关于connect by的性能分析详见http://www.os2ora.com/connect-by-performance-tunning/ [...] [...] 最后从http://www.os2ora.com/how-to-return-continuous-dates-between-two-dates/发现一个同样的案例,OS2提供了一个非常不错的方法：关于connect by的性能分析详见http://www.os2ora.com/connect-by-performance-tunning/ [...]

]]> By: robinma http://www.os2ora.com/connect-by-performance-tunning/comment-page-1/#comment-666 robinma Fri, 16 Jul 2010 09:41:31 +0000 http://www.os2ora.com/connect-by-performance-tunning/#comment-666 好文章 好文章

]]> By: admin http://www.os2ora.com/connect-by-performance-tunning/comment-page-1/#comment-11 admin Sat, 18 Apr 2009 09:49:02 +0000 http://www.os2ora.com/connect-by-performance-tunning/#comment-11 嗯，一般情况下，碰到这种问题时，很多人可能有类似的想法，就是用dual来实现。例如为了实现生成一个连续的整数列表，如1,2,3...100，可以简单的： <pre lang="sql" line="1"> select level from dual connect by level<100; </pre> 这里当然可以用这种思路，加上一个with 子句就得到了你提到的这种写法。 不过性能我觉得是不相上下的，你说的性能比前两个好不知是基于什么标准？执行时间？执行计划？ 从执行计划看，我其实还是更倾向于我原来的第二种写法: <pre lang="sql" line="1"> SQL> set autot trace exp SQL> SELECT begin_date + level - 1 one_date 2 FROM (SELECT min(begin_date) begin_date, max(end_date) end_date 3 FROM test) test 4 connect BY begin_date + level - 1 < = end_date 5 ; Elapsed: 00:00:00.00 Execution Plan ---------------------------------------------------------- Plan hash value: 928680046 ------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | ------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 1 | 18 | 3 (0)| 00:00:01 | |* 1 | CONNECT BY WITHOUT FILTERING| | | | | | | 2 | VIEW | | 1 | 18 | 3 (0)| 00:00:01 | | 3 | SORT AGGREGATE | | 1 | 18 | | | | 4 | TABLE ACCESS STORAGE FULL| TEST | 2 | 36 | 3 (0)| 00:00:01 | ------------------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 1 - filter("END_DATE">=INTERNAL_FUNCTION("BEGIN_DATE")+LEVEL-1) Note ----- - dynamic sampling used for this statement SQL> with abc as 2 (SELECT /*+ materialize */ max(end_date) ed,min(begin_date) bd ,max(end_date)-min(begin_date) days FROM test) 3 select (select bd from abc)+level from dual connect by level<(select days from abc); Elapsed: 00:00:00.00 Execution Plan ---------------------------------------------------------- Plan hash value: 2352707181 ------------------------------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | ------------------------------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 1 | | 5 (0)| 00:00:01 | | 1 | VIEW | | 1 | 9 | 2 (0)| 00:00:01 | | 2 | TABLE ACCESS FULL | SYS_TEMP_0FD9D6602_21C39D13 | 1 | 18 | 2 (0)| 00:00:01 | | 3 | TEMP TABLE TRANSFORMATION | | | | | | | 4 | LOAD AS SELECT | DUAL | | | | | | 5 | SORT AGGREGATE | | 1 | 18 | | | | 6 | TABLE ACCESS STORAGE FULL | TEST | 2 | 36 | 3 (0)| 00:00:01 | |* 7 | CONNECT BY WITHOUT FILTERING| | | | | | | 8 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 | | 9 | VIEW | | 1 | 6 | 2 (0)| 00:00:01 | | 10 | TABLE ACCESS FULL | SYS_TEMP_0FD9D6602_21C39D13 | 1 | 18 | 2 (0)| 00:00:01 | ------------------------------------------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 7 - filter(LEVEL< (SELECT "DAYS" FROM (SELECT /*+ CACHE_TEMP_TABLE ("T1") */ "C0" "ED","C1" "BD",INTERNAL_FUNCTION("C2") "DAYS" FROM "SYS"."SYS_TEMP_0FD9D6602_21C39D13" "T1") "ABC")) Note ----- - dynamic sampling used for this statement SQL> </pre> 嗯，一般情况下，碰到这种问题时，很多人可能有类似的想法，就是用dual来实现。例如为了实现生成一个连续的整数列表，如1,2,3…100，可以简单的：

1

SELECT level FROM dual connect BY level<100;

这里当然可以用这种思路，加上一个with 子句就得到了你提到的这种写法。

不过性能我觉得是不相上下的，你说的性能比前两个好不知是基于什么标准？执行时间？执行计划？

从执行计划看，我其实还是更倾向于我原来的第二种写法:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67

SQL> SET autot trace exp
SQL> SELECT begin_date + level - 1 one_date
  2  FROM (SELECT min(begin_date) begin_date, max(end_date) end_date
  3  FROM test) test
  4  connect BY begin_date + level - 1 < = end_date
  5  ;
Elapsed: 00:00:00.00
 
Execution Plan
----------------------------------------------------------
Plan hash value: 928680046
 
-------------------------------------------------------------------------------------
| Id  | Operation                    | Name | Rows  | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT             |      |     1 |    18 |     3   (0)| 00:00:01 |
|*  1 |  CONNECT BY WITHOUT FILTERING|      |       |       |            |          |
|   2 |   VIEW                       |      |     1 |    18 |     3   (0)| 00:00:01 |
|   3 |    SORT AGGREGATE            |      |     1 |    18 |            |          |
|   4 |     TABLE ACCESS STORAGE FULL| TEST |     2 |    36 |     3   (0)| 00:00:01 |
-------------------------------------------------------------------------------------
 
Predicate Information (IDENTIFIED BY operation id):
---------------------------------------------------
 
   1 - filter("END_DATE">=INTERNAL_FUNCTION("BEGIN_DATE")+LEVEL-1)
 
Note
-----
   - dynamic sampling used FOR this statement
 
SQL> WITH abc AS
  2  (SELECT /*+ materialize */ max(end_date) ed,min(begin_date) bd ,max(end_date)-min(begin_date) days FROM test)
  3  SELECT (SELECT bd FROM abc)+level FROM dual connect BY level<(SELECT days FROM abc);
Elapsed: 00:00:00.00
 
Execution Plan
----------------------------------------------------------
Plan hash value: 2352707181
 
-------------------------------------------------------------------------------------------------------------
| Id  | Operation                     | Name                        | Rows  | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT              |                             |     1 |       |     5   (0)| 00:00:01 |
|   1 |  VIEW                         |                             |     1 |     9 |     2   (0)| 00:00:01 |
|   2 |   TABLE ACCESS FULL           | SYS_TEMP_0FD9D6602_21C39D13 |     1 |    18 |     2   (0)| 00:00:01 |
|   3 |  TEMP TABLE TRANSFORMATION    |                             |       |       |            |          |
|   4 |   LOAD AS SELECT              | DUAL                        |       |       |            |          |
|   5 |    SORT AGGREGATE             |                             |     1 |    18 |            |          |
|   6 |     TABLE ACCESS STORAGE FULL | TEST                        |     2 |    36 |     3   (0)| 00:00:01 |
|*  7 |   CONNECT BY WITHOUT FILTERING|                             |       |       |            |          |
|   8 |    FAST DUAL                  |                             |     1 |       |     2   (0)| 00:00:01 |
|   9 |    VIEW                       |                             |     1 |     6 |     2   (0)| 00:00:01 |
|  10 |     TABLE ACCESS FULL         | SYS_TEMP_0FD9D6602_21C39D13 |     1 |    18 |     2   (0)| 00:00:01 |
-------------------------------------------------------------------------------------------------------------
 
Predicate Information (IDENTIFIED BY operation id):
---------------------------------------------------
 
   7 - filter(LEVEL< (SELECT "DAYS" FROM  (SELECT /*+ CACHE_TEMP_TABLE ("T1") */ "C0" "ED","C1"
              "BD",INTERNAL_FUNCTION("C2") "DAYS" FROM "SYS"."SYS_TEMP_0FD9D6602_21C39D13" "T1") "ABC"))
 
Note
-----
   - dynamic sampling used FOR this statement
 
SQL>
]]> By: zhang yong hui http://www.os2ora.com/connect-by-performance-tunning/comment-page-1/#comment-10 zhang yong hui Fri, 17 Apr 2009 09:31:04 +0000 http://www.os2ora.com/connect-by-performance-tunning/#comment-10 你好， 看了你这篇文章很有启发， 我想出了另一种sql的写法： with abc as (SELECT /*+ materialize */ max(end_date) ed,min(begin_date) bd ,max(end_date)-min(begin_date) days FROM test) select (select bd from abc)+level from dual connect by level<(select days from abc); 经过测试， 这个sql的性能比前两个都要好。 你好， 看了你这篇文章很有启发， 我想出了另一种sql的写法：
with abc as
(SELECT /*+ materialize */ max(end_date) ed,min(begin_date) bd ,max(end_date)-min(begin_date) days FROM test)
select (select bd from abc)+level from dual connect by level<(select days from abc);

经过测试， 这个sql的性能比前两个都要好。

]]> By: Yefeng http://www.os2ora.com/connect-by-performance-tunning/comment-page-1/#comment-2 Yefeng Mon, 23 Mar 2009 13:50:09 +0000 http://www.os2ora.com/connect-by-performance-tunning/#comment-2 谢谢，受教了 谢谢，受教了

]]>