为什么不走INDEX FAST FULL SCAN呢

 INDEX FULL SCAN 索引全扫描。单块读 。它扫描的结果是有序的，因为索引是有序的。它通常发生在
                                        下面几种情况(注意:即使SQL满足以下情况 不一定会走索引全扫描)
                                       1. SQL语句有order by选项，并且order by 的列都包含
                                        在索引中，并且order by 后列顺序必须和索引列顺序一致。
                                       2. 在进行SORT MERGE JOIN的时候，如果要查询的列通过索
                                          引就能获得，那就不必进行全表扫描了，另外也避免了排
                                          序，因为INDEX FULL SCAN返回的结果已经排序。
                                       3. 当查询中有GROUP BY，并且GROUP BY 的列包含在索引中。
                               等待事件：db file sequential read
                               HINT: INDEX(表名/别名 索引名)
   INDEX FAST FULL SCAN 索引快速全扫描。多块读 。当SQL要查询的数据能够完全从索引中获得，那么
                                                 Oracle就不会走全表扫描了，就会走索引快速全
                                                 扫描。索引快速全扫描类似全表扫描，它可以多块
                                                 读，并且可以并行扫描。
                               等待事件：db file scattered read
                               HINT：INDEX_FFS(表名/别名 索引名) 

测试INDEX FAST FULL SCAN:
create table test as select * from dba_objects;

create index test_idx1 on test(object_name);

BEGIN
  DBMS_STATS.GATHER_TABLE_STATS(ownname          => 'HR',
                                tabname          => 'TEST',
                                estimate_percent => 30,
                                method_opt       => 'for all columns size repeat',
                                no_invalidate    => FALSE,
                                degree           => 8,
                                cascade          => TRUE);
END;

SQL>  select object_name from test where object_name is not null and owner='SYS';

Execution Plan
----------------------------------------------------------
Plan hash value: 1357081020

--------------------------------------------------------------------------
| Id  | Operation	  | Name | Rows  | Bytes | Cost (%CPU)| Time	 |
--------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |	 | 40056 |  1212K|  4689   (1)| 00:00:57 |
|*  1 |  TABLE ACCESS FULL| TEST | 40056 |  1212K|  4689   (1)| 00:00:57 |
--------------------------------------------------------------------------

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

   1 - filter("OWNER"='SYS' AND "OBJECT_NAME" IS NOT NULL)

为什么不走INDEX FAST FULL SCAN呢？

因为test_idx1只存放了(object_name)的数据,但是没有存放OWNER的数据,索引无法过滤OWENR='SYS'部分数据,只能访问表来过滤。

create index test_idx2 on test(object_name,owner)

SQL>  select object_name from test where object_name is not null and owner='SYS';

Execution Plan
----------------------------------------------------------
Plan hash value: 2636476231

----------------------------------------------------------------------------------
| Id  | Operation	     | Name	 | Rows  | Bytes | Cost (%CPU)| Time	 |
----------------------------------------------------------------------------------
|   0 | SELECT STATEMENT     |		 | 40056 |  1212K|  1822   (1)| 00:00:22 |
|*  1 |  INDEX FAST FULL SCAN| TEST_IDX2 | 40056 |  1212K|  1822   (1)| 00:00:22 |
----------------------------------------------------------------------------------

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

   1 - filter("OWNER"='SYS' AND "OBJECT_NAME" IS NOT NULL

此时走INDEX FAST FULL SCAN