Question

我有一个表，你可以在我的sql server中看到：

<div id="myId"></div>

我在这张表里面有16000000条记录。所以你知道这个表存储了所有关于cars的信息。所以我决定根据汽车模型创建文件组，这意味着我在一个文件组中保存了相同型号的汽车，你可以看到：

CREATE TABLE [dbo].[Cars](
    [Id] [bigint] IDENTITY(1,1) NOT NULL,
    [VIN] [nvarchar](max) NULL,
    [ChassisNumber] [nvarchar](max) NULL,
    [PlaqueCoded] [nvarchar](max) NULL,
    [EngineNumber] [nvarchar](max) NULL,
    [SystemType] [nvarchar](max) NULL,
    [CarType] [nvarchar](max) NULL,
    [CarTipe] [nvarchar](max) NULL,
    [FuelType] [nvarchar](max) NULL,
    [FuelSystem] [nvarchar](max) NULL,
    [Model] [int] NULL,
    [Color] [nvarchar](max) NULL,
    [SubmitDatetime] [datetime] NOT NULL,
    [ExpireDatetime] [datetime] NOT NULL,
    [ReferenceOrganization] [nvarchar](max) NULL,
    [ReferenceId] [nvarchar](max) NULL,
    [Comment] [nvarchar](max) NULL,
 CONSTRAINT [PK_Cars] PRIMARY KEY NONCLUSTERED 
(
    [Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [FullOrderDateRangePScheme]([Model])

所以我创建了这个函数

 FILEGROUP [Filegroup_1395] 
( NAME = N'data_1395', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL13.MSSQLSERVER\MSSQL\DATA\data_1395.ndf' , SIZE = 10240KB , MAXSIZE = 10240000KB , FILEGROWTH = 512000KB ), 
 FILEGROUP [Filegroup_1396] 
( NAME = N'data_1396', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL13.MSSQLSERVER\MSSQL\DATA\data_1396.ndf' , SIZE = 10240KB , MAXSIZE = 10240000KB , FILEGROWTH = 512000KB ), 
 FILEGROUP [Filegroup_1397] 
( NAME = N'data_1397', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL13.MSSQLSERVER\MSSQL\DATA\data_1397.ndf' , SIZE = 566976KB , MAXSIZE = 10240000KB , FILEGROWTH = 512000KB )

作为一个注释，VIN是唯一的。我的主要搜索是vin.i在我的桌子上创建这些索引：

CREATE PARTITION FUNCTION [FullOrderDateKeyRangePFN](int) AS RANGE LEFT FOR VALUES ( 1395, 1396, 1397)
GO
/****** Object:  PartitionScheme [FullOrderDateRangePScheme]    Script Date: 8/8/2017 11:51:38 PM ******/
CREATE PARTITION SCHEME [FullOrderDateRangePScheme] AS PARTITION [FullOrderDateKeyRangePFN] TO ( [Filegroup_1395], [Filegroup_1396], [Filegroup_1397])
GO

我的查询是：

CREATE CLUSTERED INDEX [ClusteredIndex-20170808-232559] ON [dbo].[Cars]
(
    [Model] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [FullOrderDateRangePScheme]([Model])
GO
SET ANSI_PADDING ON
GO

CREATE NONCLUSTERED INDEX [NonClusteredIndex-20170808-232524] ON [dbo].[Cars]
(
    [Id] ASC
)
INCLUDE (   [VIN]) WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [FullOrderDateRangePScheme]([Model])
GO

我的问题是：
我怎样才能做更好的表现呢？我的索引能帮忙吗？

我在调整方面是如此新颖

Answer 1

您的查询是

select * from cars where vin='IRFC1374GH7162K'

您当前的索引对您的查询一点都没有帮助..

所以为了让你的查询表现更好，我建议..添加一个非聚集索引

CREATE NONCLUSTERED INDEX [NonClusteredIndex-20170808-232524] ON [dbo].[Cars]
(
    [vin] ASC
)
INCLUDE (  remaining columns)

这将是一个问题，因为如果您使用的是sql2016之前的版本，则最大索引键大小为900字节，对于大于等于2016的版本，限制为1700字节。

我也看到，你的搜索栏不是那么大（IRFC1374GH7162K），所以我建议调整到nvarchar（100）。所以所有列都需要，因为你正在做select *

有了这个，你将有一个单独的结构维护......但这有助于你的查询..如果你不需要所有的列，那么开销会更少

我的搜索在sql server中很慢

1 个答案: