Table name:  DE.PRODUCTS
       Index type:  SORTED
       Status:      New
       Cardinality: Table: 5265  Index: 35
       Workload:    Multipurpose

         Recommended columns
         -------------------

         QC_STATUS     --col_name
             3 references by an equality operator.  --?

         COMPLETE_STATUS  --col_name
             2 references by an equality operator.

         CREATE_BY   --col_name
             1 reference by an equality operator.

       The index type was originally set to SORTED as the default structure.
       This permits both equality and inequality segment lookups.  It also
       permits partial key and key-only-boolean lookups. --?

       ******************** SQL Statement Information ********************

       Request:     142_de_history.Req_select_0005

       Importance:  Specified: 5000     Calculated: 2472.80  --?
       Executions:  Collected: 2        Calculated: 2   --?
       Disk reads:  Collected: 492      Average:    246.00 --?

       The calculated importance was derived from the average disk reads pe
      
      
tuning report 建议我见一个索引
  CREATE INDEX DE.PRODUCTS_IDX_003
     ON DE.PRODUCTS
        (QC_STATUS,
         COMPLETE_STATUS,
         CREATE_BY)
    TABLESPACE DE
    PCTFREE 0
    STORAGE (
        INITIAL 512K
        NEXT 512K
        PCTINCREASE 0
        MINEXTENTS 1
        MAXEXTENTS 1024);

谁能帮我解释解释,尤其带? 的

没有操作时:
$ sar -qu 5 5

SunOS sprk250-3 5.8 Generic_108528-09 sun4u 08/20/01

22:46:45 runq-sz %runocc swpq-sz %swpocc
%usr %sys %wio %idle

22:46:50
0 3 0 97

22:46:55
1 1 0 97

22:47:00
0 3 0 96

22:47:05 1.0 20
0 1 1 98

22:47:10
0 1 0 98


Average 1.0 4
Average 0 2 0 97
正执行一条查询:
$ sar -qu 5 5

SunOS sprk250-3 5.8 Generic_108528-09 sun4u 08/20/01

22:48:58 runq-sz %runocc swpq-sz %swpocc
%usr %sys %wio %idle

22:49:03 1.0 20
53 9 5 33

22:49:08
45 7 2 47

22:49:13 1.0 20
46 7 1 45

22:49:18 1.0 20
53 10 0 37

22:49:23
64 10 1 26


Average 1.0 12
Average 52 9 2 37
$

1)定期 分析索引, 以便确定是否需要重建

rebuilding index when following:
1. height > 4 or
2. del_lf_rows/lf_rows > 25% or
3. lf_rows is small while lf_blks is large


2) 定期分析表, 使ORACLE 在执行计划时会用到索引
execute dbms_stats.gather_schema_stats('schema_name');
or
analyze table table_name compute statistics;

还有什么?

tuning pack的有关索引的分析, 大概是做的最糟糕的一部分了, 看不懂也罢!

个人认为, 它给出的建议很少有有价值的.

我只是把它作为一个筛子,  筛出一批嫌疑名单, 对名单中的索引还要自己判断.

它建议新建的索引大多时候也是没有必要的. 不过, 与上面相同, 凡是这中间涉及的表, 值得重点分析一下而已.

纯属个人意见.