第一次oracle sql tuning，简直就是乱来，见笑了
很好的参考：
http://www.iteye.com/topic/498902
1.首先在ADDM中看到建议调优的sql，跑一下SQL Tuning Advisor
看到一些建议，在EM这里更多的是建议你建index（毕竟这是最简单的方法）
2.观察需要调整表的结构，看是否适合做index
还是已经做了而因为有函数等原因使得失效
3.请程序员检查代码
正如参考里面说的，更改代码的效果往往要比建index要好得多
以下是具体的优化结果，
其实就是在原有基础上缩减了日期范围(acct_date已经做了index)，使得子集大大减少（重180天到1天）
原来:
优化后：
写在最后：
觉得SQL tuning的东西DBA的组织力、执行力比纯技术要重要得多！！
http://www.iteye.com/topic/498902
1.首先在ADDM中看到建议调优的sql，跑一下SQL Tuning Advisor
看到一些建议，在EM这里更多的是建议你建index（毕竟这是最简单的方法）
2.观察需要调整表的结构，看是否适合做index
还是已经做了而因为有函数等原因使得失效
ALTER INDEX index_name MONITORING USAGE; SELECT * FROM V$OBJECT_USAGE;
3.请程序员检查代码
正如参考里面说的，更改代码的效果往往要比建index要好得多
以下是具体的优化结果，
其实就是在原有基础上缩减了日期范围(acct_date已经做了index)，使得子集大大减少（重180天到1天）
原来:
SQL> SET SERVEROUTPUT ON;
SQL> SET LINESIZE 1000;
SQL> SET AUTOTRACE TRACEONLY;
SQL> SELECT q.work_ctr, SUBSTR (q.prod_nbr, 2, 2) AS size1,
2 q.prod_nbr AS prod_nbr, COALESCE (a.pass_qty, 0) pass_qty,
3 q.qty AS inspect_qty
4 FROM (SELECT main_wc AS work_ctr, prod_nbr, SUM (pass_qty) pass_qty
5 FROM HMASPAS_SOUTPUT
6 WHERE fac_id = 'A'
7 AND trans_date >= '2009-11-24 07.30.00'
8 AND trans_date <= '2009-11-24 19.30.00'
9 GROUP BY main_wc, prod_nbr) a
10 RIGHT JOIN
11 (SELECT work_ctr, prod_nbr, COUNT (*) AS qty
12 FROM HMASHIS_WQDH
13 WHERE fac_id = 'A'
14 AND trans_date >= '2009-11-24 07.30.00'
15 AND trans_date <= '2009-11-24 19.30.00'
16 GROUP BY work_ctr, prod_nbr) q
17 ON q.work_ctr = a.work_ctr AND q.prod_nbr = a.prod_nbr
18 ORDER BY q.work_ctr, q.prod_nbr, a.pass_qty, q.qty;
37 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 544019557
---------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 56 | 85722 (2)| 00:17:09 |
| 1 | SORT ORDER BY | | 1 | 56 | 85722 (2)| 00:17:09 |
|* 2 | HASH JOIN OUTER | | 1 | 56 | 85721 (2)| 00:17:09 |
| 3 | VIEW | | 1 | 28 | 5 (20)| 00:00:01 |
| 4 | HASH GROUP BY | | 1 | 44 | 5 (20)| 00:00:01 |
|* 5 | TABLE ACCESS BY INDEX ROWID| HMASHIS_WQDH | 1 | 44 | 4 (0)| 00:00:01 |
|* 6 | INDEX RANGE SCAN | INDEX2_WQDH | 1 | | 3 (0)| 00:00:01 |
| 7 | VIEW | | 29 | 812 | 85715 (2)| 00:17:09 |
| 8 | HASH GROUP BY | | 29 | 1363 | 85715 (2)| 00:17:09 |
|* 9 | TABLE ACCESS FULL | HMASPAS_SOUTPUT | 29 | 1363 | 85714 (2)| 00:17:09 |
---------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("Q"."PROD_NBR"="A"."PROD_NBR"(+) AND "Q"."WORK_CTR"="A"."WORK_CTR"(+))
5 - filter("FAC_ID"='A')
6 - access("TRANS_DATE">='2009-11-24 07.30.00' AND "TRANS_DATE"<='2009-11-24 19.30.00')
9 - filter("TRANS_DATE">='2009-11-24 07.30.00' AND "FAC_ID"='A' AND
"TRANS_DATE"<='2009-11-24 19.30.00')
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
393981 consistent gets
324974 physical reads
0 redo size
2043 bytes sent via SQL*Net to client
509 bytes received via SQL*Net from client
4 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
37 rows processed
优化后：
SQL> SET SERVEROUTPUT ON;
SQL> SET LINESIZE 1000;
SQL> SET AUTOTRACE TRACEONLY;
SELECT q.work_ctr, SUBSTR (q.prod_nbr, 2, 2) AS size1,
SQL> 2 q.prod_nbr AS prod_nbr, COALESCE (A.pass_qty, 0) pass_qty,
3 q.qty AS inspect_qty
4 FROM (SELECT main_wc AS work_ctr, prod_nbr, SUM (pass_qty) pass_qty
5 FROM HMASPAS_SOUTPUT
6 WHERE fac_id = 'A'
7 AND acct_date >=
8 TO_CHAR (TO_DATE ('2009-11-24', 'YYYY-MM-DD') - 1,
9 'YYYY-MM-DD'
10 )
11 AND acct_date <=
12 TO_CHAR (TO_DATE ('2009-11-24', 'YYYY-MM-DD') + 1,
13 'YYYY-MM-DD'
14 )
15 AND trans_date >= '2009-11-24 07.30.00'
16 AND trans_date <= '2009-11-24 19.30.00'
17 GROUP BY main_wc, prod_nbr) A
18 RIGHT JOIN
19 (SELECT work_ctr, prod_nbr, COUNT (*) AS qty
20 FROM HMASHIS_WQDH
21 WHERE fac_id = 'A'
22 AND acct_date >=
23 TO_CHAR (TO_DATE ('2009-11-24', 'YYYY-MM-DD') - 1,
24 'YYYY-MM-DD'
25 )
26 AND acct_date <=
27 TO_CHAR (TO_DATE ('2009-11-24', 'YYYY-MM-DD') + 1,
28 'YYYY-MM-DD'
29 )
30 AND trans_date >= '2009-11-24 07.30.00'
31 AND trans_date <= '2009-11-24 19.30.00'
32 GROUP BY work_ctr, prod_nbr) q
33 ON q.work_ctr = A.work_ctr AND q.prod_nbr = A.prod_nbr
34 ORDER BY q.work_ctr, q.prod_nbr, A.pass_qty, q.qty;
37 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 2416477161
-------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 56 | 1010 (1)| 00:00:13 |
| 1 | SORT ORDER BY | | 1 | 56 | 1010 (1)| 00:00:13 |
|* 2 | HASH JOIN OUTER | | 1 | 56 | 1009 (1)| 00:00:13 |
| 3 | VIEW | | 1 | 28 | 5 (20)| 00:00:01 |
| 4 | HASH GROUP BY | | 1 | 55 | 5 (20)| 00:00:01 |
|* 5 | TABLE ACCESS BY INDEX ROWID| HMASHIS_WQDH | 1 | 55 | 4 (0)| 00:00:01 |
|* 6 | INDEX RANGE SCAN | INDEX2_WQDH | 1 | | 3 (0)| 00:00:01 |
| 7 | VIEW | | 1 | 28 | 1003 (1)| 00:00:13 |
| 8 | HASH GROUP BY | | 1 | 58 | 1003 (1)| 00:00:13 |
|* 9 | TABLE ACCESS BY INDEX ROWID| HMASPAS_SOUTPUT | 1 | 58 | 1002 (1)| 00:00:13 |
|* 10 | INDEX RANGE SCAN | IDX_HMASPAS_SOUTPUT | 31455 | | 102 (1)| 00:00:02 |
-------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("Q"."PROD_NBR"="A"."PROD_NBR"(+) AND "Q"."WORK_CTR"="A"."WORK_CTR"(+))
5 - filter("ACCT_DATE">='2009-11-23' AND "FAC_ID"='A' AND "ACCT_DATE"<='2009-11-25')
6 - access("TRANS_DATE">='2009-11-24 07.30.00' AND "TRANS_DATE"<='2009-11-24 19.30.00')
9 - filter("TRANS_DATE">='2009-11-24 07.30.00' AND "FAC_ID"='A' AND
"TRANS_DATE"<='2009-11-24 19.30.00')
10 - access("ACCT_DATE">='2009-11-23' AND "ACCT_DATE"<='2009-11-25')
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
1726 consistent gets
0 physical reads
0 redo size
2043 bytes sent via SQL*Net to client
509 bytes received via SQL*Net from client
4 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
37 rows processed
写在最后：
觉得SQL tuning的东西DBA的组织力、执行力比纯技术要重要得多！！