.net批量导入数据

批量导入数据（ADO.NET)
如果你需要在程序中批量插入成千上万行的数据，你会怎么编写代码呢？最近在帮朋友调优这个的时候，总结了几种方法，并对其进行比较。

大概的界面如下，我模拟了一个客户资料表.

 

数据我是放在一个XML文件的，大约6734行。类似下面的格式

<?xml version="1.0" encoding="utf-8" ?>
<root>
<Customers>
  <CustomerID>ALFKI</CustomerID>
  <CompanyName>Sina</CompanyName>
  <ContactName>Maria Anders</ContactName>
  <ContactTitle>Sales Representative</ContactTitle>
  <Address>Obere Str. 57</Address>
  <City>Berlin</City>
  <PostalCode>12209</PostalCode>
  <Country>Germany</Country>
  <Phone>030-0074321</Phone>
  <Fax>030-0076545</Fax>
</Customers>
<Customers>
  <CustomerID>ANATR</CustomerID>
  <CompanyName>Ana Trujillo Emparedados y helados</CompanyName>
  <ContactName>Ana Trujillo</ContactName>
  <ContactTitle>Owner</ContactTitle>
  <Address>Avda. de la Constitución 2222</Address>
  <City>México D.F.</City>
  <PostalCode>05021</PostalCode>
  <Country>Mexico</Country>
  <Phone>(5) 555-4729</Phone>
  <Fax>(5) 555-3745</Fax>
</Customers>
<Customers>
  <CustomerID>ANTON</CustomerID>
  <CompanyName>Antonio Moreno Taquería</CompanyName>
  <ContactName>Antonio Moreno</ContactName>
  <ContactTitle>Owner</ContactTitle>
  <Address>Mataderos  2312</Address>
  <City>México D.F.</City>
  <PostalCode>05023</PostalCode>
  <Country>Mexico</Country>
  <Phone>(5) 555-3932</Phone>
</Customers>
</root>下面首先在服务器稍微准备一下环境

USE [tempdb]
GO

SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
CREATE TABLE [dbo].[Customers](
    [CustomerID] [nchar](5) NOT NULL,
    [CompanyName] [nvarchar](40) NOT NULL,
    [ContactName] [nvarchar](30) NULL,
    [ContactTitle] [nvarchar](30) NULL,
    [Address] [nvarchar](60) NULL,
    [City] [nvarchar](15) NULL,
    [Region] [nvarchar](15) NULL,
    [PostalCode] [nvarchar](10) NULL,
    [Country] [nvarchar](15) NULL,
    [Phone] [nvarchar](24) NULL,
    [Fax] [nvarchar](24) NULL
) ON [PRIMARY]

 

CREATE PROCEDURE [dbo].[usp_InsertCustomer]
    @CustomerID nchar(5),
    @CompanyName nvarchar(40),
    @ContactName nvarchar(30),
    @ContactTitle nvarchar(30),
    @Address nvarchar(60),
    @City nvarchar(15),
    @Region nvarchar(15),
    @PostalCode nvarchar(10),
    @Country nvarchar(15),
    @Phone nvarchar(24),
    @Fax nvarchar(24)
AS

SET NOCOUNT ON

INSERT INTO [dbo].[Customers] (
    [CustomerID],
    [CompanyName],
    [ContactName],
    [ContactTitle],
    [Address],
    [City],
    [Region],
    [PostalCode],
    [Country],
    [Phone],
    [Fax]
) VALUES (
    @CustomerID,
    @CompanyName,
    @ContactName,
    @ContactTitle,
    @Address,
    @City,
    @Region,
    @PostalCode,
    @Country,
    @Phone,
    @Fax
)  我们在tempdb中创建了一个表和一个存储过程 首先，我们把数据加载到一个DataSet        DataSet ds = new DataSet();
        private void btLoadData_Click(object sender, EventArgs e)
        {
            string dataFile = "CustomersData.xml";
            ds.ReadXml(dataFile);
            bindingSource1.DataSource = ds;
            bindingSource1.DataMember = "Customers";
            dataGridView1.DataSource = bindingSource1;
           
        }然后，我们第一个测试代码是遍历这个DataSet，每一行提交一次

        private string GetConnectionString()
        {
            return "server=(local);database=tempdb;integrated security=true;";
        }
        /// <summary>
        /// 直接遍历，一个一个的提交给服务器。时间为265毫秒左右
        /// 每一行都需要写日志
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        private void btOneByOne_Click(object sender, EventArgs e)
        {
            SqlConnection conn = new SqlConnection(GetConnectionString());
            SqlCommand cmd = conn.CreateCommand();
            cmd.CommandText = "usp_InsertCustomer";
            cmd.CommandType = CommandType.StoredProcedure;
            conn.Open();

            TimeSpan startTime = System.Diagnostics.Process.GetCurrentProcess().UserProcessorTime;

            foreach (DataRow row in ds.Tables[0].Rows)
            {
                cmd.Parameters.Clear();
                SqlParameter[] param = new SqlParameter[]{
                    new SqlParameter("@CustomerID",row[0].ToString()),
                    new SqlParameter("@CompanyName",row[1].ToString()),
                    new SqlParameter("@ContactName",row[2].ToString()),
                    new SqlParameter("@ContactTitle",row[3].ToString()),
                    new SqlParameter("@Address",row[4].ToString()),
                    new SqlParameter("@City",row[5].ToString()),
                    new SqlParameter("@Region",row[6].ToString()),
                    new SqlParameter("@PostalCode",row[7].ToString()),
                    new SqlParameter("@Country",row[8].ToString()),
                    new SqlParameter("@Phone",row[9].ToString()),
                    new SqlParameter("@Fax",row[10].ToString())
                };
                cmd.Parameters.AddRange(param);
                cmd.ExecuteNonQuery();
            }
            conn.Close();

            TimeSpan duration = System.Diagnostics.Process.GetCurrentProcess().UserProcessorTime.Subtract(startTime);
            MessageBox.Show("已经全部插入成功，所用时间为" + duration.Milliseconds.ToString() + "毫秒");
        }接下来，我们使用ADO.NET内置的一个DataAdapter来提交

        /// <summary>
        /// 这是使用Adapter的方式，其实还是遍历，而且语法也没有简单
        /// 同时，速度甚至更慢。时间为650毫秒左右
        /// 每一行都需要写日志
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        private void btUseAdapter_Click(object sender, EventArgs e)
        {
            SqlDataAdapter adapter = new SqlDataAdapter();
            adapter.AcceptChangesDuringUpdate = false;//为了演示目的，把这个开关关掉，以免它在更新完成后把数据集标记为未更改
            adapter.UpdateBatchSize = 10;//这个好像也没有什么用
           
            SqlConnection conn = new SqlConnection(GetConnectionString());
            SqlCommand insertCommand = conn.CreateCommand();
            insertCommand.CommandText = "usp_InsertCustomer";
            insertCommand.CommandType = CommandType.StoredProcedure;

            insertCommand.UpdatedRowSource = UpdateRowSource.None;

            insertCommand.Parameters.Add("@CustomerID", SqlDbType.NChar, 5, "CustomerID");
            insertCommand.Parameters.Add("@CompanyName", SqlDbType.NVarChar, 40, "CompanyName");
            insertCommand.Parameters.Add("@ContactName", SqlDbType.NVarChar, 30, "ContactName");
            insertCommand.Parameters.Add("@ContactTitle", SqlDbType.NVarChar, 30, "ContactTitle");
            insertCommand.Parameters.Add("@Address", SqlDbType.NVarChar, 60, "Address");
            insertCommand.Parameters.Add("@City", SqlDbType.NVarChar, 15, "City");
            insertCommand.Parameters.Add("@Region", SqlDbType.NVarChar, 15, "Region");
            insertCommand.Parameters.Add("@PostalCode", SqlDbType.NVarChar, 10, "PostalCode");
            insertCommand.Parameters.Add("@Country", SqlDbType.NVarChar, 15, "Country");
            insertCommand.Parameters.Add("@Phone", SqlDbType.NVarChar, 24, "Phone");
            insertCommand.Parameters.Add("@Fax", SqlDbType.NVarChar, 24, "Fax");

            adapter.InsertCommand = insertCommand;
            TimeSpan startTime = System.Diagnostics.Process.GetCurrentProcess().UserProcessorTime;
            adapter.Update(ds,"Customers");
            TimeSpan duration = System.Diagnostics.Process.GetCurrentProcess().UserProcessorTime.Subtract(startTime);
            MessageBox.Show("已经全部插入成功，所用时间为" + duration.Milliseconds.ToString() + "毫秒");

           
           
        }最后，我们找到了最快的方法

        /// <summary>
        /// 使用新的API，批量导入,这个速度很快，大约26毫秒，很显然，这种方式只写一次日志，不会为每一行写日志
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        private void btBCP_Click(object sender, EventArgs e)
        {
            using (SqlConnection conn = new SqlConnection(GetConnectionString()))
            {
                SqlBulkCopy bcp = new SqlBulkCopy(conn);
                bcp.DestinationTableName = "Customers";
                bcp.BatchSize = 100;//这是批尺寸可以调整
                for (int i = 0; i < 11; i++)
                {
                    bcp.ColumnMappings.Add(i, i);
                }
                TimeSpan startTime = System.Diagnostics.Process.GetCurrentProcess().UserProcessorTime;
                conn.Open();
                bcp.WriteToServer(ds.Tables[0]);
                TimeSpan duration = System.Diagnostics.Process.GetCurrentProcess().UserProcessorTime.Subtract(startTime);
                MessageBox.Show("已经全部插入成功，所用时间为" + duration.Milliseconds.ToString() + "毫秒");

            }
        }还有一种办法是通过在服务器OPENXML，因为XML反复处理效率很差，所以就没有测试了，可以断定它肯定比其他几种还要慢。

另外提示一下，如果不用编程的方式，那么有其他三个可能的途径去做这个事情

1. BCP工具（这是一个命令行，可以做导入和导出，不过来源文件如果不规范，那么可能很费劲）

2. BULK INSERT语句（这是一个T-SQL语句，只能做导入，我们上面使用的SQLBULKCopy应该和他很类似）

3. XML Bulk Load（这是一套COM的对象模型，适合导入XML文档）

还有，在做大量的数据导入和导出时，可以考虑微软为SQL Server配套的SSIS（Integration Service)