everyone:
Great. A small requirement is realized, and three ways of writing are used. The execution plan is as follows:
SQL> SELECT DAY_ID,NAME,URL,CNT
2 FROM (SELECT DAY_ID,NAME,URL,CNT,row_number() OVER(PARTITION BY DAY_ID, NAME ORDER BY CNT DESC) RANK
3 FROM (SELECT T.DAY_ID,
4 T.NAME,
5 T.URL,
6 COUNT(1) CNT
7 FROM TEST T
8 GROUP BY T.DAY_ID, T.NAME, T.URL) d)
9 WHERE RANK=1;
执行计划
----------------------------------------------------------
Plan hash value: 481755404
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 22 | 1144 | 5 (40)| 00:00:01 |
|* 1 | VIEW | | 22 | 1144 | 5 (40)| 00:00:01 |
|* 2 | WINDOW SORT PUSHED RANK| | 22 | 572 | 5 (40)| 00:00:01 |
| 3 | HASH GROUP BY | | 22 | 572 | 5 (40)| 00:00:01 |
| 4 | TABLE ACCESS FULL | TEST | 22 | 572 | 3 (0)| 00:00:01 |
---------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("RANK"=1)
2 - filter(ROW_NUMBER() OVER ( PARTITION BY "T"."DAY_ID","T"."NAME"
ORDER BY COUNT(*) DESC )<=1)
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
0 recursive calls
0 db block gets
7 consistent gets
0 physical reads
0 redo size
729 bytes sent via SQL*Net to client
415 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
5 rows processed
SQL> SELECT DAY_ID,NAME,URL,CNT
2 FROM (SELECT T.DAY_ID,
3 T.NAME,
4 T.URL,
5 COUNT(1) CNT,
6 row_number() OVER(PARTITION BY T.DAY_ID,T.NAME ORDER BY COUNT(1) DESC) RANK
7 FROM TEST T
8 GROUP BY T.DAY_ID,
9 T.NAME,
10 T.URL)
11 WHERE RANK=1;
执行计划
----------------------------------------------------------
Plan hash value: 481755404
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 22 | 1144 | 5 (40)| 00:00:01 |
|* 1 | VIEW | | 22 | 1144 | 5 (40)| 00:00:01 |
|* 2 | WINDOW SORT PUSHED RANK| | 22 | 572 | 5 (40)| 00:00:01 |
| 3 | HASH GROUP BY | | 22 | 572 | 5 (40)| 00:00:01 |
| 4 | TABLE ACCESS FULL | TEST | 22 | 572 | 3 (0)| 00:00:01 |
---------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("RANK"=1)
2 - filter(ROW_NUMBER() OVER ( PARTITION BY "T"."DAY_ID","T"."NAME"
ORDER BY COUNT(*) DESC )<=1)
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
0 recursive calls
0 db block gets
7 consistent gets
0 physical reads
0 redo size
729 bytes sent via SQL*Net to client
415 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
5 rows processed
SQL> SELECT DAY_ID,NAME,URL,cnt
2 FROM (SELECT T.DAY_ID,
3 T.NAME,
4 T.URL,
5 COUNT(1) cnt,
6 MAX(COUNT(1)) OVER(PARTITION BY DAY_ID,NAME) cnt1
7 FROM TEST T
8 GROUP BY T.DAY_ID, T.NAME, T.URL) d
9 WHERE cnt=cnt1;
执行计划
----------------------------------------------------------
Plan hash value: 1133560901
-----------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-----------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 22 | 1144 | 4 (25)| 00:00:01 |
|* 1 | VIEW | | 22 | 1144 | 4 (25)| 00:00:01 |
| 2 | WINDOW BUFFER | | 22 | 572 | 4 (25)| 00:00:01 |
| 3 | SORT GROUP BY | | 22 | 572 | 4 (25)| 00:00:01 |
| 4 | TABLE ACCESS FULL| TEST | 22 | 572 | 3 (0)| 00:00:01 |
-----------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("CNT"="CNT1")
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
0 recursive calls
0 db block gets
7 consistent gets
0 physical reads
0 redo size
729 bytes sent via SQL*Net to client
415 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
2 sorts (memory)
0 sorts (disk)
5 rows processed
SQL> spool off;
It can be seen that the first type of writing is essentially the same as the second type. The third type of writing is more sorted in memory, and the effect is a bit poor.