Question

我有这样的东西：

With a as (select id from table_a)
select * from table_b where table_b.id > (select min(id) from a)

table_a是一个巨大的表，具有数百万条记录，我不想每次使用min（i）时都要遍历所有记录来查找min（i）。有什么办法可以将min（id）存储在变量中并在查询中使用它？像这样：

With a as (select id from table_a),
b as ((select min(id) into min_id from a))
select * from table_b where table_b.id > min_id

Answer 1

您的查询应该在做您想要的。但是，如果您确实希望确定，请将子查询移至from子句：

select b.*
from table_b b join
     (select min(id) as min_id from a) a
     on b.id > a.min_id;

Answer 2

您可以通过在列table_a上的id上定义索引来帮助Oracle，而不用遍历所有记录来查找最小值。

如果定义了索引，Oracle将执行一次索引访问以获取最小行。

没有索引，您将执行大表的FULL TABLE SCAN个。

验证的最佳方法是遵守执行计划：

 create index idx_a on table_a(id);  

EXPLAIN PLAN  SET STATEMENT_ID = 'q1' into   plan_table  FOR
With a as (select id from table_a)
select * from table_b 
where table_b.id > (select min(id) from a);
SELECT * FROM table(DBMS_XPLAN.DISPLAY('plan_table', 'q1','ALL')); 

---------------------------------------------------------------------------------------
| Id  | Operation                   | Name    | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |         |  5524 | 71812 |    49   (3)| 00:00:01 |
|*  1 |  TABLE ACCESS FULL          | TABLE_B |  5524 | 71812 |    47   (3)| 00:00:01 |
|   2 |   SORT AGGREGATE            |         |     1 |    13 |            |          |
|   3 |    INDEX FULL SCAN (MIN/MAX)| IDX_A   |     1 |    13 |     2   (0)| 00:00:01 |
---------------------------------------------------------------------------------------

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

   1 - filter("TABLE_B"."ID"> (SELECT MIN("ID") FROM "TABLE_A" "TABLE_A"))

Oracle在SQL查询中使用变量

2 个答案: