oracle的表類型

Oracle表分類：普通堆表，索引組織表，簇表，全局臨時表，外部表，分區表。

普通堆表的不足之處：
1.表更新日誌開銷(能夠在適當的場合選擇全局臨時表)
2.表的delete操做有瑕疵(可在適當場合考慮全局臨時表和分區表)
3.表的數據太大檢索慢(可在適當場合選擇分區表)
4.索引回表讀開銷大(可在適當的場合選擇索引組織表)
5.有序插入難有序讀出,(後插入的記錄首先尋找並填充 delete 操做後的 "空塊") (可在適當的場合選擇簇表)

 回表：在數據中，當查詢數據的時候，在索引中查找索引後，得到該行的rowid，根據rowid再查詢表中數據，就是回表。
create table s1 as select * from dba_objects where rownum<=200;
create index index_obj on s1(object_id);

set autotrace traceonly;

LISN@orcl>select * from s1 where object_id < 10;
select * from s1 where object_id < 10;

8 rows selected.


Execution Plan
----------------------------------------------------------
Plan hash value: 3647141056

-----------------------------------------------------------------------------------------
| Id  | Operation            | Name    | Rows    | Bytes | Cost (%CPU)| Time    |
-----------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT        |        |     8 |  1656 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| S1    |     8 |  1656 |     2   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN        | INDEX_OBJ |     8 |    |     1   (0)| 00:00:01 |
-----------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("OBJECT_ID"<10)

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
      0  recursive calls
      0  db block gets
      4  consistent gets
      0  physical reads
      0  redo size
       2235  bytes sent via SQL*Net to client
    523  bytes received via SQL*Net from client
      2  SQL*Net roundtrips to/from client
      0  sorts (memory)
      0  sorts (disk)
      8  rows processed


-- INDEX RANGE SCAN| INDEX_OBJ    不須要回表


#####################