Question

我对查询优化的知识不多，在这里我需要一些帮助。这是我们项目的示例代码，它是长期运行的oracle查询，用于从2个不同的表（student，student_info）中提取数据。

是否有可能对此进行优化？ where子句“和”操作在这里如何工作？

在执行AND子句时，它保持任何顺序吗？下面的查询如何通过删除行b.student_id in ('a123','b123','c123')的代码来使前后有所不同。

我们没有权限在该表列上添加索引。

如何在不创建索引的情况下提高性能。

select a.student_id
       max(decode(a.marks_limit, 99.99,100,null )) as max_marks,
       b.student_city_code "NYC",
from student a, 
     student_info b
where a.student_id=b.student_id
  and a.student_id in ('a123','b123','c123')
  and b.student_id in ('a123','b123','c123')
  and  b.adress_modified > TO_TIMESTAMP('2003/12/13 10:13:18', 'YYYY/MM/DD HH:MI:SS')
  group by a.student_id, b.student_city_code;

Answer 1

您可以：

select a.student_id
    max(decode(a.marks_limit, 99.99,100,null )) as max_marks,
    b.student_city_code "NYC",
from student a
join student_info b
  on a.student_id=b.student_id     -- explicit join
where a.student_id in ('a123','b123','c123') -- removing duplicate condition
  and b.adress_modified>TO_TIMESTAMP('2003/12/13 10:13:18','YYYY/MM/DD HH:MI:SS')
group by a.student_id, b.student_city_code;

并添加索引：

CREATE INDEX id1 ON student(student_id);
CREATE INDEX id2 ON student_info(student_id);
CREATE INDEX id3 ON student_info(adress_modified);

Answer 2

首先，使用具有合理的表别名和JOIN表达式的正确，明确，正确 case语法编写查询。

修复后，我会期望像这样：

select s.student_id
       max(s.marks_limit) as max_marks,  -- I have no idea what the decode() is supposed to be doing
       si.student_city_code
from student s join
     student_info si
     on s.student_id = si.student_id
where s.student_id in ('a123', 'b123', 'c123')
 and   si.adress_modified > TIMESTAMP '2003-12-13T10:13:18'HH:MI:SS')
group by s.student_id, si.student_city_code;

我将从student(student_id)和student_info(student_id address_modified, student_city_code)的索引开始。

Answer 3

只是一些建议。

您已经拥有a.student_id = b.student_id，因此（'a123，'b123'，'c123'）中的condizione b.student_id只是一个有用的重复您还应该基于where子句

使用显式联接符号，而不是旧的隐式符号

   select a.student_id
    max(decode(a.marks_limit, 99.99,100,null )) as max_marks,
    b.student_city_code "NYC",
    from student a
    INNER JOIN  student_info b ON a.student_id= b.student_id
    WHERE 
    and a.student_id in ('a123','b123','c123')
    and  b.adress_modified > TO_TIMESTAMP('2003/12/13 10:13:18', 'YYYY/MM/DD HH:MI:SS')
    group by a.student_id, b.student_city_code

为了获得更好的性能，您应该检查表上的复合索引

 student_info ( adress_modified, student_id )

或

student_info ( adress_modified, student_id, student_city_code )

Oracle查询的查询优化

3 个答案: