SQLServer2005中利用xml类型实现批量操作的存储过程[原]

实现批量操作是常见的任务，比如批量删除一批数据，如果在应用程序端循环调用DAL.Entity.Delete则效率非常低，因为数据库往返次数取决与数据量大小。而编写一个存储过程效率更高。

做这个批量操作的存储过程使用什么作为参数是一个问题。如果使用VARCHAR(MAX)，内容是，例如要删除记录的id并以逗号隔开，例如“1,2,3,4,…..”，这样的话就会存在最大长度受限制的问题。

这儿介绍一种使用XML的VARBINARY类型作为参数的高效方法。

首先是xml的格式：







<Employees>

 <Employee EmployeeID="1" EmployeeName="Bush" EmployeeAddress="WashtonD.C." />

 <Employee EmployeeID="2" EmployeeName="Jamson" EmployeeAddress="Russia" />

 <Employee EmployeeID="3" EmployeeName="Brown" EmployeeAddress="London" />

</Employees>

 

把这个XML转换为Binary类型：

 

using System.IO;

using System.Xml;

private byte[] GetEmployeesListXmlStream(IList EmployeesList)

{

   byte[] result = null;

   using (MemoryStream stream = new MemoryStream())

   {

         XmlWriterSettings settings = new XmlWriterSettings();

         settings.Indent = true;

         settings.OmitXmlDeclaration = true;

         settings.IndentChars = "\t";

         XmlWriter writer = XmlWriter.Create(stream, settings);

         writer.WriteStartDocument();

         writer.WriteStartElement("Employees");

         foreach (Employee var in EmployeesList)

         {

             writer.WriteStartElement("Employee");

             writer.WriteAttributeString("EmployeeID", var.EmployeeID.ToString());

             writer.WriteAttributeString("EmployeeName", var.EmployeeName);

             writer.WriteAttributeString("EmployeeAddress", var.EmployeeAddress);

             writer.WriteEndElement();

         }

         writer.WriteEndElement();

         writer.WriteEndDocument();

         writer.Flush();

         writer.Close();

         result = stream.ToArray();

   }

   return result;

}

 

大家也知道，DataSet中的数据可以直接序列化成xml，通过dataSet.GetXml()，然后转化为binary即可。

 

XML的VARBINARY类型作为参数，批量删除的存储过程：

 

   1: CREATE PROCEDURE [dbo].[_Employee_BulkDelete]    

   2: (    

   3:     @EmployeeList VARBINARY(MAX)

   4: )    

   5: AS    

   6:  

   7:     --XML ----

   8:     DECLARE @EmployeeListXML XML;

   9:  

  10:     -- TABLE to store XML records ----

  11:     DECLARE @TempEmployees TABLE

  12:     (

  13:         EmployeeID BIGINT NULL,

  14:         EmployeeName NVARCHAR(512) NULL DEFAULT(''),

  15:         EmployeeAddress NVARCHAR(512) NULL DEFAULT('')

  16:     )

  17:  

  18:     -- Get XML from binary stream -----

  19:     SET @EmployeeListXML = CONVERT(XML, @EmployeeList);

  20:  

  21:     -- Fill to table-------

  22:     INSERT INTO @TempEmployees    

  23:     (    

  24:         EmployeeID,

  25:         EmployeeName,

  26:         EmployeeAddress

  27:     )    

  28:     SELECT    

  29:         nref.value('@id', 'BIGINT'),

  30:         nref.value('@name', 'NVARCHAR(512)'),

  31:         nref.value('@address', 'NVARCHAR(512)')

  32:     FROM     

  33:         @EmployeeListXML.nodes('/Employees/Employee') R(nref)

  34:  

  35:     --- Delete ----Maybe more complicated routine here !----------

  36:  

  37:     DELETE [dbo].[TempEmployees]

  38:     FROM [dbo].[TempEmployees] AS T1

  39:       INNER JOIN @TempEmployees AS T2    

  40:             ON    (T1.EmployeeID = T2.EmployeeID)    

  41:     WHERE (EmployeeSalary > 500000)

 

最后，如果想进一步优化该存储过程的效率，可以优化索引，使用临时表等。

------- 批量插入可以用SqlBulkCopy 参考：http://www.cnblogs.com/xiaoweinet/archive/2010/03/02/1676478.html