sql中order by与索引排序的区别

SQL> conn yang/yang as sysdba

已连接。

SQL> create table t as select object_id id ,object_name name

  2  from dba_objects ;

表已创建。

SQL> set autot traceonly

一次普通的全表扫描，没有排序的！

SQL> select id ,name from t;

已选择68372行。

执行计划

----------------------------------------------------------                      

Plan hash value: 1601196873                                                    

--------------------------------------------------------------------------      

| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |      

--------------------------------------------------------------------------      

|   0 | SELECT STATEMENT  |      | 64794 |  4998K|    91   (2)| 00:00:02 |      

|   1 |  TABLE ACCESS FULL| T    | 64794 |  4998K|    91   (2)| 00:00:02 |      

--------------------------------------------------------------------------      

Note                                                                            

-----                                                                           

   - dynamic sampling used for this statement

统计信息

----------------------------------------------------------                      

         68  recursive calls                                                    

          0  db block gets                                                      

       4943  consistent gets                                                    

        325  physical reads                                                     

          0  redo size                                                          

    2936793  bytes sent via SQL*Net to client                                   

      50554  bytes received via SQL*Net from client                             

       4560  SQL*Net roundtrips to/from client                                  

          0  sorts (memory)                                                     

          0  sorts (disk)                                                       

      68372  rows processed    

--根据id 排序！注意执行计划里面的TempSpc 是临时空间seo顾问,大小11M

SQL> select id ,name from t order by id;

已选择68372行。

执行计划

----------------------------------------------------------                      

Plan hash value: 961378228                                                      

----------------------------------------------------------------------------------               

| Id  | Operation          | Name | Rows  | Bytes |TempSpc| Cost (%CPU)| Time |

--------------------------------------------------------------------------------

|   0 | SELECT STATEMENT   |      | 64794 |  4998K|    |  1283   (1)| 00:00:16 |

|   1 |  SORT ORDER BY     |      | 64794 |  4998K|    11M|  1283   (1)| 00:00:16 |

|   2 |   TABLE ACCESS FULL| T    | 64794 |  4998K|       |    91   (2)| 00:00:02 |

-----------------------------------------------------------------------------------

Note                                                                            

-----                                                                           

   - dynamic sampling used for this statement                                   

统计信息

----------------------------------------------------------                      

          4  recursive calls                                                    

          0  db block gets                                                      

        394  consistent gets                                                    

          0  physical reads                                                     

          0  redo size                                                          

    2663362  bytes sent via SQL*Net to client                                   

      50554  bytes received via SQL*Net from client                             

       4560  SQL*Net roundtrips to/from client                                  

          1  sorts (memory)                                                     

          0  sorts (disk)                                                       

      68372  rows processed                                                     

--在表的 id 字段建立索引，并进行信息统计。

SQL> create index idx_id on t(id) ;

索引已创建。

SQL> exec dbms_stats.gather_table_stats(user ,'T',cascade => true);

PL/SQL 过程已成功完成。

SQL> select id ,name from t order by id;

已选择68372行。

执行计划

----------------------------------------------------------                      

Plan hash value: 961378228                                                     

-----------------------------------------------------------------------------------

| Id  | Operation          | Name | Rows  | Bytes |TempSpc| Cost (%CPU)| Time   |

-----------------------------------------------------------------------------------

|   0 | SELECT STATEMENT   |      | 68372 |  1936K|       |   638   (1)| 00:00:08 |

|   1 |  SORT ORDER BY     |      | 68372 |  1936K|  5384K|   638   (1)| 00:00:08 |

|   2 |   TABLE ACCESS FULL| T    | 68372 |  1936K|       |    91   (2)| 00:00:02 |

-----------------------------------------------------------------------------------              

统计信息

----------------------------------------------------------                      

        151  recursive calls                                                    

          0  db block gets                                                      

        348  consistent gets                                                    

          0  physical reads                                                     

          0  redo size                                                          

    2663362  bytes sent via SQL*Net to client                                   

      50554  bytes received via SQL*Net from client                             

       4560  SQL*Net roundtrips to/from client                                  

          5  sorts (memory) --没有使用索引比全表扫描多了四此排序。

          0  sorts (disk)                                                       

      68372  rows processed    

--使用索引。执行计划中没有tempspac                                              

SQL> select id ,name from t where id <1200 order by id;--加上了order by

已选择1133行。

执行计划

----------------------------------------------------------

Plan hash value: 827754323

--------------------------------------------------------------------------------------

| Id  | Operation                   | Name   | Rows  | Bytes | Cost (%CPU)| Time     |

--------------------------------------------------------------------------------------

|   0 | SELECT STATEMENT            |        |  1167 | 33843 |    10   (0)| 00:00:01 |

|   1 |  TABLE ACCESS BY INDEX ROWID| T      |  1167 | 33843 |    10   (0)| 00:00:01 |

|*  2 |   INDEX RANGE SCAN          | IDX_ID |  1167 |       |     4   (0)| 00:00:01 |

--------------------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

   2 - access("ID"<1200)

统计信息

----------------------------------------------------------

          1  recursive calls  

          0  db block gets   

        160  consistent gets

          0  physical reads  

          0  redo size   

      37872  bytes sent via SQL*Net to client  

       1241  bytes received via SQL*Net from client

         77  SQL*Net roundtrips to/from client   

          0  sorts (memory) --这里可以看出没有排序！

          0  sorts (disk)   

       1133  rows processed

小结：

如果数据直接从索引获取，也是有序的，此时加order by,cbo不会执行sort 排序动作的。即，加上order by对性能也不会有什么影响！

其实这里还是有疑问的

1 根据id 排序 走全表扫描和建立了索引后信息统计上有差别，前者比后者少了3个sort 操作！而后者的TempSpc比全表扫描少了将近一半！

2 关于TempSpc 的理解如果是临时表空间 ，就用到了磁盘排序了 ，而执行上面没有显示disk sort！