Question

根据我在Wikipedia上阅读的内容，相关的子查询在性能方面并不是最好的，因为每次主查询中的每条记录都需要执行子查询。

由于这个原因，我试图重写以下查询。我尝试将查询重写为仅使用INNER JOIN来获得不同的结果。

有人可以帮我弄清楚我在新查询中错过了什么吗？另外，我正在使用MSSQL server 2008。

原始查询

SELECT * 
FROM 
    table1 t
WHERE 
    t.some_Id=123
AND EXISTS
(
    SELECT TOP  1 1       
    FROM 
        table2 t2 WITH(NOLOCK)
    LEFT OUTER JOIN 
        table3 t3 WITH(NOLOCK)     
            ON  t3.another_id = t2.another_id 
                AND t3.yet_another_id = table1.yet_another_id //correlation with table1  
        WHERE t2.t2_id = 456  AND   t3.t3_id IS NULL
)

新查询

SELECT * 
FROM 
    table1 t
WHERE 
    t.some_Id=123
AND EXISTS
(
    SELECT TOP  1 1       
    FROM 
        table2 t2 WITH(NOLOCK)
    LEFT OUTER JOIN 
        table3 t3 WITH(NOLOCK)     
            ON  t3.another_id = t2.another_id
        INNER JOIN table1 t1
            ON t3.yet_another_id = t1.yet_another_id
        WHERE t2.t2_id = 456  AND   t3.t3_id IS NULL
)

Answer 1

这是您的查询：

SELECT * 
FROM table1 t
WHERE t.some_Id = 123 AND
      EXISTS (SELECT TOP  1 1       
              FROM table2 t2 LEFT OUTER JOIN 
                   table3 t3    
                   ON  t3.another_id = t2.another_id 
              WHERE t2.t2_id = 456  AND t3.t3_id IS NULL AND
                    t3.yet_another_id = table1.yet_another_id
             );

如果要提高性能，请考虑以下索引：

table1(some_id, yet_another_id)
table2(t2_id, another_id)
table3(another_id, yet_another_id, t3_id)

重写相关子查询

1 个答案: