Question

select distinct m1.jobreference,m1.assid,m2.assid 
From dbo.mess m1
inner join dbo.mess m2 
    on m1.JobReference=m2.jobreference and m1.assid!=m2.assid
where 
m1.jobreference not in ( 
select Jobreference from dbo.mess where routingkey='j.asschanged')

4958行受影响

表＆＃39>表＆＃39;混乱＆＃39;扫描计数20，逻辑读取359816，物理读取525，预读读取4626，lob逻辑读取0，lob物理读取0，lob预读读取0。表＆＃39;工作文件＆＃39;。扫描计数0，逻辑读取0，物理读取0，预读取读取0，lob逻辑读取0，lob物理读取0，lob预读读取0。表＆＃39;工作台＆＃39;。扫描计数0，逻辑读取0，物理读取0，预读读取0，lob逻辑读取0，lob物理读取0，lob预读读取0. 强文本

Answer 1

Sub-Query不需要。试试这种方式

SELECT m1.jobreference,
       m1.assid,
       m2.assid
FROM   dbo.mess m1
       INNER JOIN PRD_Phoenix.dbo.mess m2
               ON m1.JobReference = m2.jobreference
                  AND m1.assid != m2.assid
Group By  m1.jobreference,m1.assid,m2.assid
Having count(case when m1.routingkey = 'j.asschanged' then 1 end) = 0

如果您确实需要删除重复项，请使用Distinct。如果您的查询不会产生重复，或者您不担心重复，请从查询中删除DISTINCT关键字

Answer 2

我首先将查询编写为：

select distinct jobreference, assid, assid2
from (select m1.jobreference, m1.assid, m2.assid as assid2,
             sum(case when m2.routingkey = 'j.asschanged' then 1 else 0 end) over (partition by m2.jobreference) as cnt
      from dbo.mess m1 inner join
           PRD_Phoenix.dbo.mess m2 
           on m1.JobReference = m2.jobreference and m1.assid <> m2.assid
     ) mm
where cnt = 0;

然后，您要考虑明显的索引是m2(jobreference, assid)和m1(jobference, assid)。

我担心select distinct和高基数连接的可能性。如果没有关于数据结构的更多信息，很难提出进一步的建议。

优化查询作为其慢人

2 个答案: