如果子句之间有重叠的数据，Oracle 会展平多个子句吗？在我的应用程序中，用户可以创建动态搜索词，因此可能存在重叠数据。 Oracle 会为我优化 SQL，还是我必须在创建 SQL 之前计算它？
即
从 xxtable 中选择 id WHERE id (BETWEEN 10 AND 20) or (BETWEEN 18 AND 30)
它会“按原样”运行还是转换为：
从 xxtable 中选择 id WHERE id (BETWEEN 10 AND 30)
感谢您的宝贵时间。
【问题讨论】：
当我们对带有索引的真实表进行试验时，值得看看会发生什么。此示例表有 69,241 行，在 COL_3 上有一个非唯一索引，带有统计信息。
案例 1：基本的两个重叠的 BETWEEN 子句
SQL> set autotrace traceonly exp
SQL>
SQL> select * from big_table
2 where col_3 between 0.8 and 1
3 or col_3 between 0.9 and 1.1
4 /
Execution Plan
----------------------------------------------------------
Plan hash value: 3993303771
-------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 14737 | 805K| 176 (1)| 00:00:03 |
|* 1 | TABLE ACCESS FULL| BIG_TABLE | 14737 | 805K| 176 (1)| 00:00:03 |
-------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("COL_3"<=1.1 AND "COL_3">=0.9 OR "COL_3"<=1 AND
"COL_3">=0.8)
SQL>
结果：索引被忽略，然后进行全表扫描
案例 2： BETWEEN 子句共享一个上限
SQL> select * from big_table
2 where col_3 between 0.8 and 1.1
3 or col_3 between 0.9 and 1.1
4 /
Execution Plan
----------------------------------------------------------
Plan hash value: 1461639892
-----------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-----------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 7924 | 433K| 114 (0)| 00:00:02 |
| 1 | TABLE ACCESS BY INDEX ROWID| BIG_TABLE | 7924 | 433K| 114 (0)| 00:00:02 |
|* 2 | INDEX RANGE SCAN | BIG3_IDX | 7924 | | 17 (0)| 00:00:01 |
-----------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("COL_3"<=1.1)
filter("COL_3">=0.8 OR "COL_3">=0.9)
SQL>
结果：索引用于上限，避免了全表扫描
案例 3： BETWEEN 子句共享一个下限
SQL> select * from big_table
2 where col_3 between 0.8 and 1.1
3 or col_3 between 0.8 and 1.2
4 /
Execution Plan
----------------------------------------------------------
Plan hash value: 3993303771
-------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 15146 | 828K| 176 (1)| 00:00:03 |
|* 1 | TABLE ACCESS FULL| BIG_TABLE | 15146 | 828K| 176 (1)| 00:00:03 |
-------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter(("COL_3"<=1.2 OR "COL_3"<=1.1) AND "COL_3">=0.8)
SQL>
结果：索引被忽略，然后进行全表扫描
案例 4：两个 BETWEEN 范围合并为一个子句
SQL> select * from big_table
2 where col_3 between 0.8 and 1.1
3 /
Execution Plan
----------------------------------------------------------
Plan hash value: 1461639892
-----------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-----------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 7924 | 433K| 114 (0)| 00:00:02 |
| 1 | TABLE ACCESS BY INDEX ROWID| BIG_TABLE | 7924 | 433K| 114 (0)| 00:00:02 |
|* 2 | INDEX RANGE SCAN | BIG3_IDX | 7924 | | 17 (0)| 00:00:01 |
-----------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("COL_3">=0.8 AND "COL_3"<=1.1)
SQL>
结果：索引同时用于上限和下限
因此，总而言之，如果两个 BETWEEN 子句重叠并且列上有索引，那么合并它们可能是值得的。
【讨论】：
这取决于优化器将如何处理它的查询。您可以使用自动跟踪在 SQLPlus 中测试它并查看谓词信息（这是在 Oracle 版本 10.2.0.3 上完成的）：
SQL>set autot traceonly
SQL>
1 select l
2 from (SELECT l
3 FROM (SELECT LEVEL l
4 FROM dual CONNECT BY LEVEL < 20
5 )
6 )
7 where l between 5 and 10
8* or l between 7 and 15;
11 rows selected.
Elapsed: 00:00:00.21
Execution Plan
----------------------------------------------------------
Plan hash value: 2403765415
--------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 13 | 2 (0)| 00:00:01 |
|* 1 | VIEW | | 1 | 13 | 2 (0)| 00:00:01 |
|* 2 | CONNECT BY WITHOUT FILTERING| | | | | |
| 3 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
--------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("L">=5 AND "L"<=10 OR "L">=7 AND "L"<=15)
2 - filter(LEVEL<20)
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
0 consistent gets
0 physical reads
0 redo size
494 bytes sent via SQL*Net to client
403 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
11 rows processed
在这种情况下，优化器不会对 where 条件进行重写，但如果我们稍微改变一下：
SQL>
1 select l
2 from (SELECT l
3 FROM (SELECT LEVEL l
4 FROM dual CONNECT BY LEVEL < 20
5 )
6 )
7 where l between 5 and 10
8* or l between 7 and 10;
6 rows selected.
Elapsed: 00:00:00.20
Execution Plan
----------------------------------------------------------
Plan hash value: 2403765415
--------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 13 | 2 (0)| 00:00:01 |
|* 1 | VIEW | | 1 | 13 | 2 (0)| 00:00:01 |
|* 2 | CONNECT BY WITHOUT FILTERING| | | | | |
| 3 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
--------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("L"<=10 AND ("L">=5 OR "L">=7))
2 - filter(LEVEL<20)
Statistics
----------------------------------------------------------
0 recursive calls
0 db block gets
0 consistent gets
0 physical reads
0 redo size
388 bytes sent via SQL*Net to client
392 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
6 rows processed
在这一点上，我们看到优化器确实认识到两个 where 条件具有相同的上限。所以它取决于查询优化器将如何重写它。
【讨论】：