SQL Server 2008中使用FileStream存取大文件

SQL Server 2008中引入了Filestream，使用它可以将非机构化大型数据（如文本文档、图像和视频）等以varbinary(max)的形式存储在文件系统中。使用数据库的备份还原功能可以将这些数据一起备份还原。详细可以查看 MSDN:http://msdn.microsoft.com/zh-cn/library/bb895234.aspx

本文将简单总结如何创建可以使用FileStream的数据库以及如何使用c#访问存取数据：

在开始之前，首先要启用FileStream，在MSDN中有详细的介绍，此处不赘述。

1. 创建数据库

创建数据时可以执行以下TSQL语句，

CREATE DATABASE TestDB
ON
PRIMARY ( NAME = TestDB,
FILENAME = 'c:\data\TestDB.mdf'),
FILEGROUP FileStreamGroup1 CONTAINS FILESTREAM( NAME = Arch3,
FILENAME = 'c:\data\filestream1')
LOG ON  ( NAME = TestDBlog,
FILENAME = 'c:\data\TestDBlog.ldf')
GO

如果是在已经创建好的数据库上启用FileStream，可以

a. Right click the “TestDB” database and select “Properties”.

b. First create a FileGroup, click “Filegroups” menu and create one under “Stream” section, named “FileStreamGroup1”. And then click “Files” menu, and add a file named “filestream1” and Set it’s FileGroup to “FileStreamGroup1”, and then specify a folder
to hold data, like c:\Data. Click ok, and you can check the folder “c:\data” to see what is added.

2. 创建表

CREATE TABLE [dbo].[TestTable2](
[ID] [uniqueidentifier] ROWGUIDCOL  NOT NULL,
[ContentType] [nvarchar](50) NOT NULL,
[FileName] [nvarchar](50) NOT NULL,
[FileContent] [varbinary](max) FILESTREAM  NULL,
[FileSize] [int] NULL,
CONSTRAINT [PK_TestTable2] PRIMARY KEY CLUSTERED
(
[ID] ASC
)WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY] FILESTREAM_ON [FileStreamGroup1]
) ON [PRIMARY] FILESTREAM_ON [FileStreamGroup1]

GO

SET ANSI_PADDING OFF
GO

如果表已经创建，可以执行以下TSQL来设置：

ALTER TABLE [dbo].[TestTable] alter column ID [uniqueidentifier] not null

ALTER TABLE [dbo].[TestTable] alter column ID add ROWGUIDCOL

ALTER TABLE [dbo].[TestTable] add FileContent varbinary(MAX) FILESTREAM;

3. 使用C#读写

读：

SqlConnection conn = null;conn = new SqlConnection(connect);
conn.Open();
tx = conn.BeginTransaction();var qry = "SELECT FileName, FileContent.PathName() as FilePath, ContentType, GET_FILESTREAM_TRANSACTION_CONTEXT() as TranContext  FROM TestTable2 WHERE ID = @ID";
var cmd = new SqlCommand(qry, conn, tx);
cmd.Parameters.AddWithValue("@ID", id);using (rdr = cmd.ExecuteReader())
{     if (rdr.HasRows)
{
rdr.Read();
fileModel = new FileModel();
fileModel.FileName = Convert.ToString(rdr["FileName"]);
fileModel.FilePath = Convert.ToString(rdr["FilePath"]);
byte[] tranContext = (byte[])rdr["TranContext"];
fileModel.ContentType = Convert.ToString(rdr["ContentType"]); ;
fileModel.FileStream = new SqlFileStream(fileModel.FilePath, tranContext, FileAccess.Read)；

}
}

得到文件的Stream，就可以对该文件进行读写操作了。

写：

public static void PostFileToDB(HttpPostedFileBase file)
{
string fileName = Path.GetFileName(file.FileName);
string contentType = file.ContentType;
int fileSize = file.ContentLength / 1024;

using (SqlConnection conn = new SqlConnection(connect))
{
conn.Open();
using (SqlTransaction trn = conn.BeginTransaction ())
{
SqlCommand cmdInsert = new SqlCommand(
@"insert into TestTable2
(FileName, FileContent, ContentType, FileSize)
output
INSERTED.FileContent.PathName(),
GET_FILESTREAM_TRANSACTION_CONTEXT ()
values
(@FileName, 0x, @ContentType, @FileSize)", conn, trn);
cmdInsert.Parameters.Add("@FileName", SqlDbType.VarChar, 256);
cmdInsert.Parameters["@FileName"].Value = fileName;

cmdInsert.Parameters.Add("@ContentType", SqlDbType.VarChar, 256);
cmdInsert.Parameters["@ContentType"].Value = contentType;                    cmdInsert.Parameters.Add("@FileSize", SqlDbType.Int);
cmdInsert.Parameters["@FileSize"].Value = fileSize;                    string path = null;
byte[] context = null;                    // cmdInsert is an INSERT command that uses the OUTPUT clause
// Thus we use the ExecuteReader to get the
// result set from the output columns
//
using (SqlDataReader rdr = cmdInsert.ExecuteReader())
{
rdr.Read();
path = rdr.GetString(0);
context = rdr.GetSqlBytes(1).Buffer;
}                    using (SqlFileStream sfs = new SqlFileStream(
path, context, FileAccess.Write))
{
file.InputStream.CopyTo(sfs);
}
trn.Commit ();
}
}
}