国内最全IT社区平台 联系我们 | 收藏本站
华晨云阿里云优惠2
您当前位置：首页 > 数据库 > 数据库应用 > 灌入大量数据后手工采集统计信息的重要性

灌入大量数据后手工采集统计信息的重要性

来源：程序员人生   发布时间：2015-01-06 08:47:40 阅读次数：2761次

1. 创建测试表TBL_STAT，及索引，但不插入记录

SQL> create table TBL_STAT as select * from dba_objects where 1<>1;
Table created.

SQL> create index idx_tbl_stat on tbl_stat (object_id);
Index created.

SQL> select count(*) from tbl_stat;
  COUNT(*)
----------
         0

2. 检索TBL_STAT的履行计划
SQL> explain plan for select object_name from tbl_stat where object_id = 1;
Explained.

SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 2448091186
------------------------------------------------------------------------------
| Id  | Operation         | Name     | Rows  | Bytes | Cost (%CPU)| Time     |
------------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |          |     1 |    79 |     2   (0)| 00:00:01 |
|*  1 |  TABLE ACCESS FULL| TBL_STAT |     1 |    79 |     2   (0)| 00:00:01 |
------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
   1 - filter("OBJECT_ID"=1)
Note
-----
   - dynamic sampling used for this statement
17 rows selected.
发现依照索引字段查询使用的是全表扫描。

3. 手工搜集TBL_STAT表的统计信息
SQL> exec dbms_stats.gather_table_stats(ownname=>'DCSOPEN', tabname=>'TBL_STAT',  estimate_percent=>100);
PL/SQL procedure successfully completed.

4. 再次检索TBL_STAT表
SQL> explain plan for select object_name from tbl_stat where object_id = 1;
Explained.

SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 3529113932
--------------------------------------------------------------------------------------------
| Id  | Operation                   | Name         | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |              |     1 |    79 |     1   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| TBL_STAT     |     1 |    79 |     1   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | IDX_TBL_STAT |     1 |       |     1   (0)| 00:00:01 |
--------------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
   2 - access("OBJECT_ID"=1)
14 rows selected.
发现这次用到了索引范围扫描，说明搜集统计信息让Oracle可以选择正确的履行计划路径。

5. 插入100万的测试记录
SQL> begin
  2    for i in 1 .. 10 loop
  3      insert into tbl_stat select * from dba_objects;
  4      commit;
  5    end loop;
  6  end;
  7  /
PL/SQL procedure successfully completed.

SQL> select count(*) from tbl_stat;
  COUNT(*)
----------
   1190725

6. 查看检索TBL_STAT表的履行计划
SQL> explain plan for select object_name from tbl_stat where object_id = 1;
Explained.

SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 3529113932
--------------------------------------------------------------------------------------------
| Id  | Operation                   | Name         | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |              |     1 |    79 |     1   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| TBL_STAT     |     1 |    79 |     1   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | IDX_TBL_STAT |     1 |       |     1   (0)| 00:00:01 |
--------------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
   2 - access("OBJECT_ID"=1)
14 rows selected.
插入100万记录后，发现还是索引范围扫描。

7. 创建第2个测试表TBL_STAT_2，和索引
SQL> create table tbl_stat_2 as select * from tbl_stat;
Table created.

SQL> create index idx_tbl_stat_2 on tbl_stat_2 (object_id);
Index created.

SQL> select count(*) from tbl_stat_2;
  COUNT(*)
----------
   1190725

8. 检索TBL_STAT和TBL_STAT_2关联查询的履行计划
SQL> explain plan for select a.object_name, b.object_name from tbl_stat a, tbl_stat_2 b where a.object_Id = b.object_id;
Explained.

SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 752230886
----------------------------------------------------------------------------------------------
| Id  | Operation                   | Name           | Rows  | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |                |     1 |   158 |    27   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| TBL_STAT_2     |    25 |  1975 |    25   (0)| 00:00:01 |
|   2 |   NESTED LOOPS              |                |     1 |   158 |    27   (0)| 00:00:01 |
|   3 |    TABLE ACCESS FULL        | TBL_STAT       |     1 |    79 |     2   (0)| 00:00:01 |
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|*  4 |    INDEX RANGE SCAN         | IDX_TBL_STAT_2 |    25 |       |     2   (0)| 00:00:01 |
----------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
   4 - access("A"."OBJECT_ID"="B"."OBJECT_ID")
Note
-----
   - dynamic sampling used for this statement
20 rows selected.
可以看到这里对TBl_STAT使用的是全表扫描，对TBL_STAT_2使用的是索引扫描，表之间是嵌套循环连接。

SQL> explain plan for select a.object_name, b.object_name from tbl_stat_2 a, tbl_stat b where a.object_Id = b.object_id;
Explained.

SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 752230886
----------------------------------------------------------------------------------------------
| Id  | Operation                   | Name           | Rows  | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |                |     1 |   158 |    27   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| TBL_STAT_2     |    25 |  1975 |    25   (0)| 00:00:01 |
|   2 |   NESTED LOOPS              |                |     1 |   158 |    27   (0)| 00:00:01 |
|   3 |    TABLE ACCESS FULL        | TBL_STAT       |     1 |    79 |     2   (0)| 00:00:01 |
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|*  4 |    INDEX RANGE SCAN         | IDX_TBL_STAT_2 |    25 |       |     2   (0)| 00:00:01 |
----------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
   4 - access("A"."OBJECT_ID"="B"."OBJECT_ID")
Note
-----
   - dynamic sampling used for this statement
20 rows selected.
即便置换两个表的连接顺序，照旧选择TBL_STAT表是全表扫描，TBL_STAT_2是索引范围扫描，但由于插入记录后未收集过统计信息，两张表的预估记录数现在都是和实际相差较多。

9. 手工收集TBL_STAT的统计信息
SQL> exec dbms_stats.gather_table_stats(ownname=>'DCSOPEN', tabname=>'TBL_STAT',  estimate_percent=>100);
PL/SQL procedure successfully completed.

SQL> explain plan for select a.object_name, b.object_name from tbl_stat_2 a, tbl_stat b where a.object_Id = b.object_id;
Explained.

SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 1789047457
-----------------------------------------------------------------------------------------
| Id  | Operation          | Name       | Rows  | Bytes |TempSpc| Cost (%CPU)| Time     |
-----------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |            |    29M|  3038M|       | 15552   (2)| 00:03:07 |
|*  1 |  HASH JOIN         |            |    29M|  3038M|    47M| 15552   (2)| 00:03:07 |
|   2 |   TABLE ACCESS FULL| TBL_STAT   |  1190K|    34M|       |  3790   (1)| 00:00:46 |
|   3 |   TABLE ACCESS FULL| TBL_STAT_2 |  1299K|    97M|       |  3645   (1)| 00:00:44 |
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - access("A"."OBJECT_ID"="B"."OBJECT_ID")
Note
PLAN_TABLE_OUTPUT
-------------------------------------------------------------------------------------
   - dynamic sampling used for this statement
19 rows selected.
发现此时TBL_STAT和TBL_STAT_2的预估行数已不是1了，而且表之间采取的是全表扫描的哈希连接。

10. 手工收集TBL_STAT_2表的统计信息
SQL> exec dbms_stats.gather_table_stats(ownname=>'DCSOPEN', tabname=>'TBL_STAT_2',  estimate_percent=>100);
PL/SQL procedure successfully completed.

SQL> explain plan for select a.object_name, b.object_name from tbl_stat_2 a, tbl_stat b where a.object_Id = b.object_id;
Explained.

SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 2620555949
-----------------------------------------------------------------------------------------
| Id  | Operation          | Name       | Rows  | Bytes |TempSpc| Cost (%CPU)| Time     |
-----------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |            |    29M|  1703M|       | 12327   (2)| 00:02:28 |
|*  1 |  HASH JOIN         |            |    29M|  1703M|    47M| 12327   (2)| 00:02:28 |
|   2 |   TABLE ACCESS FULL| TBL_STAT_2 |  1190K|    34M|       |  3644   (1)| 00:00:44 |
|   3 |   TABLE ACCESS FULL| TBL_STAT   |  1190K|    34M|       |  3790   (1)| 00:00:46 |
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - access("A"."OBJECT_ID"="B"."OBJECT_ID")
15 rows selected.
此时TBL_STAT_2表的记录也趋于和实际1致，两表的连接还是哈希连接。

总结：
1. 表的统计信息搜集还是比较重要的1项工作，除Oracle 10g以后会有自动搜集的作业外，也能够手工进行统计信息的搜集。
2. 本例中，由于TBL_STAT表灌入100万数据后，未搜集统计信息，和TBL_STAT_2表连接采取的是嵌套循环连接，这类连接适用于大表和小表的关联场景，但实际这的两张表数据量相当，且都超过了100万，这样相当于100万*100万次关联，当搜集统计信息后，两表连接改成了哈希连接，说明此时Oracle已知道了表的实际数据量，履行计划也是根据表的实际数据量来做的判断，因此当表灌入大量数据后，建议手工收集统计信息，否则在系统自动收集统计信息之前，可能得到的履行计划就是错的。

生活不易，码农辛苦
如果您觉得本网站对您的学习有所帮助,可以手机扫描二维码进行捐赠
程序员人生
------分隔线----------------------------

上一篇 用WebCollector 2.x爬取新浪微博（无需手动获取cookie)

下一篇 J2SE学习笔记（1）――递归函数

分享到:
------分隔线----------------------------
为码而活
为码而活
积分：4237
15粉丝
7关注
栏目热点

程序员人生，我编程，我富裕，记住wfuyu网，php教程，php学习，php手册，CMS模版制作

声明：本站大部分内容是作者原创，少部分收集于互联网供大家一起学习，原版权很多不明，如有侵权请联系本站，谢谢！

粤ICP备14040726号-1   2015-2020 程序员人生 版权所有

平平安安 备案网站
关闭
程序员人生