Database 有多少复合索引比单一索引慢

我正在考虑删除索引（A）并创建复合索引（A，B），因为A和B通常在where子句中一起使用

然而，我想知道如果where子句中只请求一个查询，它会对查询产生什么影响

你知道它在不同的数据库上是如何工作的吗

---

（索引是使用默认类型在例如1000万个记录表上创建的）

最好的方法是实际操作：）

所以我创建了一个表：

CREATE TABLE TestTable (ID INT,
FirstName VARCHAR(50),
LastName VARCHAR(50),
City VARCHAR(50))

然后插入100 xxx行：

INSERT INTO TestTable (ID,FirstName,LastName,City)
SELECT TOP 100000 ROW_NUMBER() OVER (ORDER BY a.name) RowID,
'Bob1',
CASE WHEN ROW_NUMBER() OVER (ORDER BY a.name)%2 = 1 THEN 'Smith2'
ELSE 'Brown3' END,
CASE WHEN ROW_NUMBER() OVER (ORDER BY a.name)%10 = 1 THEN 'New York4'
WHEN ROW_NUMBER() OVER (ORDER BY a.name)%10 = 5 THEN 'San Marino5'
WHEN ROW_NUMBER() OVER (ORDER BY a.name)%10 = 3 THEN 'Los Angeles6'
ELSE 'Houston7' END
FROM sys.all_objects a
CROSS JOIN sys.all_objects b

然后我创建了两个单独的索引：

CREATE INDEX IX_FirstName ON TestTable(FirstName)
GO
CREATE INDEX IX_LastName ON TestTable(LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

CREATE INDEX IX_FirstName_LastName
ON TestTable(FirstName, LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

查询：

CREATE INDEX IX_FirstName ON TestTable(FirstName)
GO
CREATE INDEX IX_LastName ON TestTable(LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

CREATE INDEX IX_FirstName_LastName
ON TestTable(FirstName, LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

具有两个单独索引的执行计划：

一个综合指数：

CREATE INDEX IX_FirstName ON TestTable(FirstName)
GO
CREATE INDEX IX_LastName ON TestTable(LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

CREATE INDEX IX_FirstName_LastName
ON TestTable(FirstName, LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

查询：

CREATE INDEX IX_FirstName ON TestTable(FirstName)
GO
CREATE INDEX IX_LastName ON TestTable(LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

CREATE INDEX IX_FirstName_LastName
ON TestTable(FirstName, LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

具有一个复合索引的执行计划：

结论：

CREATE INDEX IX_FirstName ON TestTable(FirstName)
GO
CREATE INDEX IX_LastName ON TestTable(LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

CREATE INDEX IX_FirstName_LastName
ON TestTable(FirstName, LastName)
GO

SELECT FirstName,
     LastName,       
     COUNT(*) RowsCount
FROM dbo.TestTable
GROUP BY FirstName, LastName, City

这实际上取决于许多因素，例如数据库管理系统、数据库管理系统的版本。然而，可以看出，两个单独的指数和一个复合指数之间没有区别

请注意，如果要使用复合索引，则应在查询中使用复合索引的所有列