从本地数据库更新远程数据库时的性能问题

Question

1. 我尝试通过触发器从本地数据库更新远程表（dblink）。 我的成本约为 24990。表数据数接近 3032965。

UPDATE remote_table
SET ip_stats = (
  CASE 
    WHEN :NEW.ip_stats IN (22,30,)  THEN 'Non-Network'
    WHEN :NEW.ip_stats IN (23,31,182)  THEN 'Cancel'
    WHEN :NEW.ip_stats IN (24,32)  THEN 'file'
    WHEN :NEW.ip_stats IN (26,34)  THEN 'online'
    WHEN :NEW.ip_stats IN (63)  THEN 'decrease'
    WHEN :NEW.ip_stats IN (64)  THEN 'increase' 
    END) ,
  ip_comments = (
  CASE :NEW.ip_commts
    WHEN 154 THEN 'increase'
    WHEN 155 THEN 'previous'
    WHEN 156 THEN 'hardware'
    WHEN 157 THEN 'software'
  END) ,
  feedback           = :NEW.feedback ,
  changed_by         = :NEW.changed_by ,
  create_dt          = :NEW.create_dt ,
  status             = 'Y' ,
  sum_amount         = NVL (:NEW.net_amount ,0)
WHERE     ID     = :NEW.new_key;

以下是更新语句的解释计划。

---------------------------------------------------------------------------------------------------
| Id  | Operation              | Name            | Rows  | Bytes | Cost (%CPU)| Time     | Inst   |
---------------------------------------------------------------------------------------------------
|   0 | UPDATE STATEMENT REMOTE|                 |     1 |   195 | 24990   (2)| 00:05:00 |        |
|   1 |  UPDATE                | remote_table    |       |       |            |          | rmt    |
|*  2 |   TABLE ACCESS FULL    | remote_table    |     1 |   195 | 24990   (2)| 00:05:00 | rmt    |
---------------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
 
   2 - filter(TO_NUMBER("A1"."ID")=123478)
 
Note
-----
   - fully remote statement

1. 此外，当我对该 remote_table 使用 count(*) 时，它的成本与 24990 相同。

注意：我在下面的选择语句中使用INDEX 提示，但它不起作用

SELECT /*+INDEX (a Column_index_name)*/ COUNT(*) FROM remote_table a WHERE a.ID = 123478;

下面是 select count(*) 语句的解释计划。

---------------------------------------------------------------------------------------------------
| Id  | Operation              | Name            | Rows  | Bytes | Cost (%CPU)| Time     | Inst   |
---------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT REMOTE|                 |     1 |     9 | 24990   (2)| 00:05:00 |        |
|   1 |  SORT AGGREGATE        |                 |     1 |     9 |            |          |        |
|*  2 |   TABLE ACCESS FULL    | remote_table    |     1 |     9 | 24990   (2)| 00:05:00 | rmt    |
---------------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
 
   2 - filter(TO_NUMBER("A1"."ID")=123478)
 
Note
-----
   - fully remote statement

任何人都可以就如何处理这个问题提出建议吗？

Answer 1

请注意列格式。你的专栏ID是VARCAHR2，所以你应该这样查询它。

您的谓词ID = 123478 进行了一个+隐式转换*，它禁止索引访问

使用ID = '123478'，它将启用索引访问。

带有id varchar2(10)的模拟表示例

 SELECT * FROM tab a WHERE a.ID = 23478;

导致执行计划（为简洁起见）

|*  1 |  TABLE ACCESS FULL| TAB  |  1081 |  2120K| 27171   (1)| 00:00:02 |

   1 - filter(TO_NUMBER("A"."ID")=23478)

但是使用正确格式文字的查询

 SELECT * FROM tab a WHERE a.ID = '23478';

有正确的计划

-----------------------------------------------------------------------------------------------
| Id  | Operation                           | Name    | Rows  | Bytes | Cost (%CPU)| Time     |
-----------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT                    |         |     1 |  2009 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID BATCHED| TAB     |     1 |  2009 |     2   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN                  | TAB_IDX |     1 |       |     1   (0)| 00:00:01 |
-----------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
 
   2 - access("A"."ID"='23478')