一、EntityFramework
EF是传统的ORM框架，也是一个比较重量级的ORM框架。这里仍然使用EF的原因在于为了突出轻量级ORM框架的性能，所谓有对比才有更优的选择。
1.1 准备一张数据库表
（1）For MSSQL
CREATE TABLE [dbo].[Posts]
(
[Id] INT NOT NULL PRIMARY KEY IDENTITY,
[CategoryId] INT NOT NULL,
[Slug] VARCHAR(120) NOT NULL,
[Title] NVARCHAR(100) NOT NULL,
[Published] DATETIME NOT NULL,
[Excerpt] NVARCHAR(MAX) NOT NULL,
[Content] NVARCHAR(MAX) NOT NULL
);
（2）For MySQL
CREATE TABLE Posts
(
Id INT NOT NULL PRIMARY KEY AUTO_INCREMENT,
CategoryId INT NOT NULL,
Slug VARCHAR(120) NOT NULL,
Title NVARCHAR(100) NOT NULL,
Published DATETIME NOT NULL,
Excerpt LONGTEXT NOT NULL,
Content LONGTEXT NOT NULL
);
1.2 使用Model First方式创建数据模型
（1）通过nuget添加EF组件引用，然后创建edmx数据模型
（2）由于EF首次使用存在效率问题，因此采用园子里推荐的EF暖机操作作为测试首次执行的代码
static void WarmupEntityFramework()
{
// EF暖机操作
using (var db = new MyAppDBContext())
{
var objectContext = ((IObjectContextAdapter)db).ObjectContext;
var mappingCollection = (System.Data.Entity.Core.Mapping.StorageMappingItemCollection)objectContext.MetadataWorkspace.GetItemCollection(System.Data.Entity.Core.Metadata.Edm.DataSpace.CSSpace);
mappingCollection.GenerateViews(new System.Collections.Generic.List<System.Data.Entity.Core.Metadata.Edm.EdmSchemaError>());
}
}
（3）写一个读取数据的方法，遍历读取Posts表记录（该表有1万行记录）
static void ModelFirstReadPosts()
{
var dbContext = new MyAppDBContext();
foreach (var item in dbContext.Posts)
{
Console.WriteLine("ID:{0},Title:{1}", item.Id, item.Title);
}
}
（4）编写入口方法，通过Stopwatch记录测试耗时
class Program
{
static Program()
{
WarmupEntityFramework();
}
static void Main(string[] args)
{
Stopwatch watch = new Stopwatch();
watch.Start();
// EF:4.9s
ModelFirstReadPosts();
watch.Stop();
Console.WriteLine("Time consumed : {0} ms", watch.ElapsedMilliseconds);
Console.ReadKey();
}
#region Method01.EntityFramework暖机操作
static void WarmupEntityFramework()
{
// EF暖机操作
using (var db = new MyAppDBContext())
{
var objectContext = ((IObjectContextAdapter)db).ObjectContext;
var mappingCollection = (System.Data.Entity.Core.Mapping.StorageMappingItemCollection)objectContext.MetadataWorkspace.GetItemCollection(System.Data.Entity.Core.Metadata.Edm.DataSpace.CSSpace);
mappingCollection.GenerateViews(new System.Collections.Generic.List<System.Data.Entity.Core.Metadata.Edm.EdmSchemaError>());
}
}
#endregion
#region Method02.Model First方式读取数据库表记录
static void ModelFirstReadPosts()
{
var dbContext = new MyAppDBContext();
foreach (var item in dbContext.Posts)
{
Console.WriteLine("ID:{0},Title:{1}", item.Id, item.Title);
}
}
#endregion
}
F5调试运行，首次查询结果如下图所示：
五次查询之后平均耗时：4.9s
二、Dapper
2.1 关于Dapper
Dapper是一个开源轻的量级的ORM，只有一个代码文件，完全开源，你可以放在项目里的任何位置，来实现数据到对象的ORM操作，体积小速度快。
2.2 使用Dapper
（1）通过nuget添加Dapper组件
（2）针对MSSQL的查询和新增操作
#region Method01.读取MSSQL单张表
// 2.7s
static void DapperReadPosts()
{
using (SqlConnection connection = new SqlConnection(connStr))
{
var postList = connection.Query<Post>("select * from Posts");
foreach (var item in postList)
{
Console.WriteLine("ID:{0},Title:{1}", item.Id, item.Title);
}
}
}
#endregion
#region Method02.读取MSSQL连接查询
// 2.6s
static void DapperReadJoin()
{
using (SqlConnection connection = new SqlConnection(connStr))
{
// 这里查询结果是动态语言类型
var postList = connection.Query("select Id,Title,GETDATE() as PostDate from Posts");
foreach (var item in postList)
{
Console.WriteLine("ID:{0},PostDate:{1}", item.Id, item.PostDate);
}
}
}
#endregion
#region Method03.读取MSSQL多个结果集
// 2.8s
static void DapperReadMultiResultSet()
{
using (SqlConnection connection = new SqlConnection(connStr))
{
using (var reader = connection.QueryMultiple("select * from Posts;select 1000 as Number;"))
{
var postList = reader.Read<Post>();
foreach (var item in postList)
{
Console.WriteLine("ID:{0},Title:{1}", item.Id, item.Title);
}
}
}
}
#endregion
#region Method04.插入MSSQL新记录
// 0.37s
static void InsertPostRecord()
{
using (SqlConnection connection = new SqlConnection(connStr))
{
// 多次插入单条记录
int count = connection.Execute("insert into Posts values(@CategoryId, @Slug, @Title, @Published, @Excerpt, @Content);", new { CategoryId = 10, Slug = "BOOK", Title = "大话设计模式", Published = DateTime.Now.AddDays(1), Excerpt = "ChengJie", Content = "Design Patterns" });
Console.WriteLine("受影响行数:{0}", count);
count = connection.Execute("insert into Posts values(@CategoryId, @Slug, @Title, @Published, @Excerpt, @Content);", new Post() { CategoryId = 10, Slug = "BOOK", Title = "大话数据结构", Published = DateTime.Now.AddDays(1), Excerpt = "ChengJie", Content = "Data Structure" });
Console.WriteLine("受影响行数:{0}", count);
// 一次插入多条记录
IList<Post> postRecords = new List<Post>();
postRecords.Add(new Post() { CategoryId = 10, Slug = "BOOK", Title = "构建之法-现代软件工程", Published = DateTime.Now.AddDays(1), Excerpt = "ZouXin", Content = "Software Engineering" });
postRecords.Add(new Post() { CategoryId = 10, Slug = "BOOK", Title = "编程之美", Published = DateTime.Now.AddDays(1), Excerpt = "ZouXin", Content = "I Love Coding" });
count = connection.Execute("insert into Posts values(@CategoryId, @Slug, @Title, @Published, @Excerpt, @Content);", postRecords);
Console.WriteLine("受影响行数:{0}", count);
}
}
#endregion
① 棒棒哒的地方1：读取多表连接查询
// 这里查询结果是动态语言类型
var postList = connection.Query("select Id,Title,GETDATE() as PostDate from Posts");
foreach (var item in postList)
{
Console.WriteLine("ID:{0},PostDate:{1}", item.Id, item.PostDate);
}
② 棒棒哒的地方2：读取多个查询结果集
using (var reader = connection.QueryMultiple("select * from Posts;select 1000 as Number;"))
{
var postList = reader.Read<Post>();
foreach (var item in postList)
{
Console.WriteLine("ID:{0},Title:{1}", item.Id, item.Title);
}
}
③ 棒棒哒的地方3：一次插入多条数据记录
// 一次插入多条记录
IList<Post> postRecords = new List<Post>();
postRecords.Add(new Post() { CategoryId = 10, Slug = "BOOK", Title = "构建之法-现代软件工程", Published = DateTime.Now.AddDays(1), Excerpt = "ZouXin", Content = "Software Engineering" });
postRecords.Add(new Post() { CategoryId = 10, Slug = "BOOK", Title = "编程之美", Published = DateTime.Now.AddDays(1), Excerpt = "ZouXin", Content = "I Love Coding" });
count = connection.Execute("insert into Posts values(@CategoryId, @Slug, @Title, @Published, @Excerpt, @Content);", postRecords);
④ 如何跨数据库读取数据记录：依赖于抽象，不依赖于具体
static void GetPostsCrossMultiDB()
{
// 依赖于抽象，不依赖于具体
using (IDbConnection connection = DbProviderFactories.GetFactory(connSetting.ProviderName).CreateConnection())
{
connection.ConnectionString = connSetting.ConnectionString;
// 使用标准SQL语句屏蔽差异
var postList = connection.Query<Post>("select * from Posts");
foreach (var item in postList)
{
Console.WriteLine("ID:{0},Title:{1}", item.Id, item.Title);
}
}
}
（3）测试结果：
| 方法 | 耗时 |
| 读取MSSQL单张表 | 2.7s |
| 读取MSSQL连接查询 | 2.6s |
| 读取MSSQL多个结果集 | 2.8s |
| 多次插入MSSQL新记录 | 148ms |
三、PetaPoco
3.1 关于PetaPoco
PetaPoco是一款适用于.NET应用程序的轻型对象关系映射器。与那些功能完备的ORM（如NHibernate或Entity Framework）不同的是，PetaPoco更注重易用性和性能，而非丰富的功能。使用PetaPoco只需要引入一个C#文件，可以使用强类型的 POCO（Plain Old CLR Object），并支持使用T4模板生成的类等等。
3.2 使用PetaPoco
（1）通过nuget添加PetaPoco组件
（2）编辑Database.tt模板文件，前提是首先将连接字符串配置正确
（3）针对MSSQL的读取和插入操作
private static void ReadAllPostData()
{
using (var context = new MyAppDBContext())
{
var postList = context.Query<Post>("select * from Posts");
foreach (var item in postList)
{
Console.WriteLine("ID:{0},Title:{1}", item.Id, item.Title);
}
}
}
private static void InsertNewPostData()
{
var post = new Post
{
CategoryId = 1,
Slug = "BOOK",
Title = "Microsoft SQL Server 2008技术内幕",
Content = Guid.NewGuid().ToString(),
Excerpt = Guid.NewGuid().ToString(),
Published = DateTime.Now
};
var count = post.Insert();
Console.WriteLine("受影响行数：{0}", count);
}
（4）测试结果：
| 方法 | 耗时 |
| 读取MSSQL单张表 | 2.7s |
| 插入MSSQL新纪录 | 30ms |
SourceCode
（1）ORMDemo：http://pan.baidu.com/s/1pJAEf0n
Reference
（1）Dapper.NET：https://github.com/StackExchange/dapper-dot-net
作者：周旭龙
出处：http://edisonchou.cnblogs.com
本文版权归作者和博客园共有，欢迎转载，但未经作者同意必须保留此段声明，且在文章页面明显位置给出原文链接。
时间： 2024-12-28 10:48:56