我正在尝试使用索引来加快处理速度，但是在进行联接的情况下，索引并不能改善查询的执行时间，在某些情况下，它会降低处理速度。
创建测试表并填充数据的查询为：
CREATE TABLE [dbo].[IndexTestTable](
[id] [int] IDENTITY(1,1) PRIMARY KEY,
[Name] [nvarchar](20) NULL,
[val1] [bigint] NULL,
[val2] [bigint] NULL)
DECLARE @counter INT;
SET @counter = 1;
WHILE @counter < 500000
BEGIN
INSERT INTO IndexTestTable
(
-- id -- this column value is auto-generated
NAME,
val1,
val2
)
VALUES
(
'Name' + CAST((@counter % 100) AS NVARCHAR),
RAND() * 10000,
RAND() * 20000
);
SET @counter = @counter + 1;
END
-- Index in question
CREATE NONCLUSTERED INDEX [IndexA] ON [dbo].[IndexTestTable]
(
[Name] ASC
)
INCLUDE ( [id],
[val1],
[val2])现在查询1，它得到了改进（只有一点点，但改进是一致的）：
SELECT *
FROM IndexTestTable I1
JOIN IndexTestTable I2
ON I1.ID = I2.ID
WHERE I1.Name = 'Name1'没有索引的统计信息和执行计划（在这种情况下，表使用默认的聚集索引）：
(5000 row(s) affected)
Table 'IndexTestTable'. Scan count 2, logical reads 5580, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
(1 row(s) affected)
SQL Server Execution Times:
CPU time = 109 ms, elapsed time = 294 ms.
现在启用索引：
(5000 row(s) affected)
Table 'IndexTestTable'. Scan count 2, logical reads 2819, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
(1 row(s) affected)
SQL Server Execution Times:
CPU time = 94 ms, elapsed time = 231 ms.
现在，由于索引的原因，查询变慢了（查询是无意义的，因为它仅用于测试）：
SELECT I1.Name,
SUM(I1.val1),
SUM(I1.val2),
MIN(I2.Name),
SUM(I2.val1),
SUM(I2.val2)
FROM IndexTestTable I1
JOIN IndexTestTable I2
ON I1.Name = I2.Name
WHERE
I2.Name = 'Name1'
GROUP BY
I1.Name启用聚集索引后：
(1 row(s) affected)
Table 'IndexTestTable'. Scan count 4, logical reads 60, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 0, logical reads 0, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 1, logical reads 155106, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
(1 row(s) affected)
SQL Server Execution Times:
CPU time = 17207 ms, elapsed time = 17337 ms.
现在禁用索引：
(1 row(s) affected)
Table 'IndexTestTable'. Scan count 5, logical reads 8642, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 2, logical reads 165212, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 0, logical reads 0, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
(1 row(s) affected)
SQL Server Execution Times:
CPU time = 17691 ms, elapsed time = 9073 ms.
问题是：
- 即使SQL Server建议使用索引，但为什么索引速度却会显着降低呢？
- 什么是大多数情况下需要使用的嵌套循环连接以及如何缩短其执行时间？
- 我做错了什么或错过了吗？
- 使用默认索引（仅在主键上），为什么要花更少的时间，而对于存在非聚集索引的连接表，则对于连接表中的每一行，应该更快地找到已连接表的行，因为连接位于“名称”列上索引已创建。这反映在查询执行计划中，并且当IndexA处于活动状态时，“索引查找”成本会降低，但是为什么还要慢一些？嵌套循环左外部联接中导致减速的原因是什么？
使用SQL Server 2012
