一、数据迁移
用来解决code first情况下当增加、删除、改变实体类,或改变DbContext类后,相应地更新数据库结构而不会对现有数据产生影响。
1、启用迁移
指令:
enable-migrations
add-migration XXX
enable-migrations指令:
a.在项目根目录下创建了一个Migrations文件夹
b.在Migrations文件夹下新建一个Configuration.cs文件。
可以通过修改Configuration.cs来对Migration做一些配置(如加入一些测试数据等)
类似初始化数据的AccountInitilizer.cs类,数据迁移的Configuration类也包含一个Seed方法。
当数据库新建或数据库结构更新后,这个方法会被调用,利用这个方法可以插入或更新test data。
2、配置Seed方法
使用drop and re-create的方式时,因为每次model改变时数据库都会被删除,所有数据都会丢失,所以需要使用DALàAccountInitilizer.cs的Seed方法来插入测试数据。
使用Code First Migrations方式,当数据库改变时测试数据会保留,所以包含test data的Seed方法一般来说是不需要的。
如果我们要部署数据库到生产环境,事实上这种情况下我们也不想Seed方法来插入测试数据到生产环境中。
但是,也有特殊情况, 比如在我们部署时获得了实际的初始化数据,如实际存在的组织部门这些初始化的信息,这时候就用到Seed方法。
与AccountInitilizer.cs类中Seed方法不同的是,以前都是用Add方法,这次用了AddOrUpdate方法来插入数据。如:
sysUsers.ForEach(s => context.SysUsers.AddOrUpdate(p => p.UserName, s));
3、执行迁移
指令:update-database
前面执行 add-migration时,同样在Migrations文件夹里面,产生一个<timestamp>_InitialCreate.cs的文件。
里面两个方法,Up和Down:
Up方法创建数据库表,Down方法删除表。
update-database指令调用了Up方法来新建database的表(和data model entity set对应), 然后调用Seed方法来填充测试数据。
数据迁移执行步骤:
1、enable-migrations(如已启用,则略过此步)
2、add-migration XXX
3、update-database
二、数据注解属性(DataAnnotations)
对某一些类型来说不需要使用Required, 例如DateTime, int,double,float,因为这些值类型不能被赋予空值,因此他们天生就具有Required的特性。
Column - 列名称
Required - not null
StringLength(20) - nvarchar(20)
TypeName - 改变SQL数据类型,如把double改成decimal
三、加载关联数据
EF有三种方式可以加载关联数据到一个实体的navigation属性中
Lazy loading(懒加载)
Eager loading(热加载)
Explicit loading(显式加载)
1、Lazy loading(懒加载)
第一次读取entity的时候不会加载。
当需要读取navigation property的时候,相关的数据将会被自动读取。
这种情况会导致多次查询数据库。
(这种加载方式是,默认不加载导航属性关联的数据,但是如果用到了,就加载。)
例如:
using (var context = new TestEntities())
{
IQueryable<Team> teams = from t in context.Teams select t;
foreach (Team t in teams)
{
Console.WriteLine(t.Players.Count());
}
Console.Read();
}
再如:
关闭懒加载方法:
(1)context.ContextOptions.LazyLoadingEnabled = false;
(2)EDMX文件属性LazyLoadingEnabled设置为False;
(3)对特定的navigation properties来说,省略property的virtual关键字就可以了(待确定);
导航属性加virtual修饰,为的是Entity Framework Code First的延迟加载功能(懒加载)。当然Entity Framework Code First延迟加载并不是必须的,所以virtual修饰符也可以不加。
优势和劣势:
当打开Lazy Loading时,我们可以不用去在意某实体是否已经加载,不会出现在调用某一实体时,出现null的尴尬,省去程序员不少心力,但同时劣势也非常明显,如果我们有大量实体,且频繁去调用相关实体,程序就会频繁地访问数据库,这很显然地会影响程序的性能。
2、Eager loading(贪婪加载)
当读取entity的时候,相关数据会被一起读取。
一般来说这种方式会产生一个join query来获取所有需要的数据。
通过Include方法来指定eager loading.
例如:
using (var context = new TestEntities())
{
var players = from p in context.Players.Include("Team").Include("PlayerDetails") where (p.PlayerDetails.Any(d => d.Height > 200)) select p;
foreach (var v in players)
{
string output = string.Format("Team: {0} Player: {1}", v.Team.TeamName, v.PlayerName);
Console.WriteLine(output);
}
Console.Read();
}
再如:
3、Explicit loading(显式加载)
和lazy loading类似,除了需要在代码中明确指定需要获取的关联数据。
在读取navigation property时explicit loading 不会自动发生,你需要手动加载相关数据。
通过获取object
state manager entry for entity,调用Collection.Load method for
collections或者Reference.Load method for properties that hold a single
entity.
一般来说,只有在关闭lazying loading的时候才会使用explicit loading.
例如:
using (TestEntities context = new TestEntities())
{
IQueryable<Player> players = from p in context.Players select p;
foreach (Player p in players)
{
if (p.Age > 30)
{
p.TeamReference.Load();
Console.WriteLine(p.PlayerName + " -> " + p.Team.TeamName);
}
}
Console.Read();
}
再如:
总结:
1、lazy loading 和 explicit loading都不立即获取property values,它们也被称作deferred loading.
2、无论使用Lazy Loading还是将Load方法放入foreach循环语句(Explicit loading),都会导致程序频繁访问数据库,导致程序性能下降。我们可以选择性地加载需要的实体。
自己总结:
1、lazy loading 和 explicit loading都会有频繁访问数据库的性能问题;所以用Eager loading(贪婪加载)比较保险点;
2、Include(Eager loading)生成其实是用的left join进行表关联,并且select所有关联的表字段(包括ID)。
下面是两个生成的SQL示例:
示例1:
var user = from a in db.sysUsers.Include("SysUserRoles").Include("SysDepartment")
where a.ID == id
select a;
生成的SQL:
SELECT
[Project1].[ID] AS [ID],
[Project1].[UserName] AS [UserName],
[Project1].[Email] AS [Email],
[Project1].[PassWord] AS [PassWord],
[Project1].[CreateDate] AS [CreateDate],
[Project1].[SysDepartmentID] AS [SysDepartmentID],
[Project1].[ID1] AS [ID1],
[Project1].[DepartmentName] AS [DepartmentName],
[Project1].[DepartmentDesc] AS [DepartmentDesc],
[Project1].[C1] AS [C1],
[Project1].[ID2] AS [ID2],
[Project1].[SysUserID] AS [SysUserID],
[Project1].[SysRoleID] AS [SysRoleID]
FROM ( SELECT
[Limit1].[ID] AS [ID],
[Limit1].[UserName] AS [UserName],
[Limit1].[Email] AS [Email],
[Limit1].[PassWord] AS [PassWord],
[Limit1].[CreateDate] AS [CreateDate],
[Limit1].[SysDepartmentID] AS [SysDepartmentID],
[Extent2].[ID] AS [ID1],
[Extent2].[DepartmentName] AS [DepartmentName],
[Extent2].[DepartmentDesc] AS [DepartmentDesc],
[Extent3].[ID] AS [ID2],
[Extent3].[SysUserID] AS [SysUserID],
[Extent3].[SysRoleID] AS [SysRoleID],
CASE WHEN ([Extent3].[ID] IS NULL) THEN CAST(NULL AS int) ELSE 1 END AS [C1]
FROM (SELECT TOP (1) [Extent1].[ID] AS [ID], [Extent1].[UserName] AS
[UserName], [Extent1].[Email] AS [Email], [Extent1].[PassWord] AS
[PassWord], [Extent1].[CreateDate] AS [CreateDate],
[Extent1].[SysDepartmentID] AS [SysDepartmentID]
FROM [dbo].[SysUser] AS [Extent1]
WHERE [Extent1].[ID] = @p__linq__0 ) AS [Limit1]
LEFT OUTER JOIN [dbo].[SysDepartment] AS [Extent2] ON [Limit1].[SysDepartmentID] = [Extent2].[ID]
LEFT OUTER JOIN [dbo].[SysUserRole] AS [Extent3] ON [Limit1].[ID] = [Extent3].[SysUserID]
) AS [Project1]
ORDER BY [Project1].[ID] ASC, [Project1].[ID1] ASC, [Project1].[C1] ASC
示例2:分页
var users = from u in db.sysUsers.Include("SysDepartment") select u;
return View(users.ToPagedList(pageNum,pageSize));
SELECT TOP (3)
[Join1].[ID1] AS [ID],
[Join1].[UserName] AS [UserName],
[Join1].[Email] AS [Email],
[Join1].[PassWord] AS [PassWord],
[Join1].[CreateDate] AS [CreateDate],
[Join1].[SysDepartmentID] AS [SysDepartmentID],
[Join1].[ID2] AS [ID1],
[Join1].[DepartmentName] AS [DepartmentName],
[Join1].[DepartmentDesc] AS [DepartmentDesc]
FROM
( SELECT [Extent1].[ID] AS [ID1], [Extent1].[UserName] AS [UserName],
[Extent1].[Email] AS [Email], [Extent1].[PassWord] AS [PassWord],
[Extent1].[CreateDate] AS [CreateDate], [Extent1].[SysDepartmentID] AS
[SysDepartmentID], [Extent2].[ID] AS [ID2], [Extent2].[DepartmentName]
AS [DepartmentName], [Extent2].[DepartmentDesc] AS [DepartmentDesc],
row_number() OVER (ORDER BY [Extent1].[UserName] ASC) AS [row_number]
FROM [dbo].[SysUser] AS [Extent1]
LEFT OUTER JOIN [dbo].[SysDepartment] AS [Extent2] ON [Extent1].[SysDepartmentID] = [Extent2].[ID]
) AS [Join1]
WHERE [Join1].[row_number] > 0
ORDER BY [Join1].[UserName] ASC
参考:
http://www.cnblogs.com/kevin2013/p/5239186.html
http://www.cnblogs.com/itjeff/p/4137733.html