Question

给出一个这样的表：

Id_Table | Id_Object_A | Id_Object_B

并且给出了ID_Object的两个子集，我必须检查这个条件：

(ID_Object_A IN (1,2,3,4,5) AND ID_Object_B IN (6,7,8,9,10))
OR
(ID_Object_A IN (6,7,8,9,10) AND ID_Object_B IN (1,2,3,4,5))

在表现方面，最好的方法是什么？

Answer 1

查询优化器应足够智能，以便最明显的解决方案具有可接受（接近最佳）的性能：

SELECT *
FROM MyTable
WHERE (ID_Object_A IN (1,2,3,4,5) AND ID_Object_B IN (6,7,8,9,10))
   OR (ID_Object_A IN (6,7,8,9,10) AND ID_Object_B IN (1,2,3,4,5))

但SQL Server的真正优势在于优化JOIN。生成一个包含10个数字的所有50种可能组合的表格：

CREATE TABLE ReferencesAB (a int NOT NULL, b int NOT NULL)
CREATE NONCLUSTERED INDEX Covering ON ReferencesAB (a, b)

INSERT INTO ReferencesAB
SELECT *
FROM (VALUES (1), (2), (3), (4), (5)) AS values1 (a)
CROSS JOIN (VALUES (6), (7), (8), (9), (10)) AS values2 (b)

INSERT INTO ReferencesAB
SELECT *
FROM (VALUES (6), (7), (8), (9), (10)) AS values1 (a)
CROSS JOIN (VALUES (1), (2), (3), (4), (5)) AS values2 (b)

然后将您的表连接到参考表：

SELECT *
FROM MyTable t
INNER JOIN ReferencesAB r ON r.a = Id_Object_A AND r.b = Id_Object_B

为了做得好，我认为最好在MyTable上对(Id_Object_A, Id_Object_B)编制索引。

尝试两种方法;研究查询计划，找出哪种方法最适合你的桌子。

查询同一对象的两个id列

1 个答案: