有个表如下：
- SQL code
CREATE TABLE [dbo].[T_FillingRec]( [FillingRec] [uniqueidentifier] NOT NULL, [Organize] [varchar](16) NULL, [SerialPort] [int] NULL, [Sequence] [bigint] NULL, [BUTime] [datetime] NULL, [BUUserName] [varchar](32) NULL, CONSTRAINT [T_Gas_FillingRec_PK] PRIMARY KEY CLUSTERED ( [FillingRec] ASC)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]) ON [PRIMARY]
其主键是[FillingRec] [uniqueidentifier] ，每次都是New一个GUID出来。
这个表是存储多个下位机传过来的数据的一个表，每个下位机占用一个串口SerialPort，每条记录的Sequence加一，但是当上位机没有给下位机应答的时候，下位机会重复传输这条记录，也就是Sequence有可能传一致的，因此这个表实际上必须根据SerialPort和Sequence作为主键是最合适的，但是表结构就是这样了，不好修改。
上位机使用了完成端口的方式处理下位机传输过来的数据，因此在同一个Sequence的记录中，可能会有不同的线程在处理数据写入的工作，可能是由于不同记录中的事务问题，导致极其偶尔的同一个串口有重复Sequence的记录存在。
请问如何解决这个问题？
现在的存储过程如下：
- SQL code
Create PROCEDURE [dbo].[pro_T_FillingRec1_Insert] @FillingRec uniqueidentifier, @Organize varchar(16), @SerialPort int, @Sequence bigint, @BUTime datetime, @BUUserName varchar(32)ASSET NOCOUNT ONIF EXISTS(SELECT * FROM [dbo].[T_FillingRec1] WHERE SerialPort = @SerialPort And Sequence = @Sequence)BEGIN UPDATE [dbo].[T_FillingRec1] SET [Organize] = @Organize, [BUTime] = @BUTime, [BUUserName] = @BUUserName WHERE SerialPort = @SerialPort And Sequence = @SequenceENDELSEBEGIN Set @FillingRec = NEWID() INSERT INTO [dbo].[T_FillingRec1] ([FillingRec],[Organize],[SerialPort],[Sequence],[BUTime],[BUUserName]) Select @FillingRec,@Organize,@SerialPort,@Sequence,@BUTime,@BUUserNameEND
请问这个存储过程有什么地方可以改进的？
------解决方案--------------------
uniqueidentifier的话无论多接近的描述，哪怕真的达到同一刻也不会重复的，你可以在表该字段直接指定默认值newid()函数，就会自动生成拉。
我的文章，希望可以帮到你：http://blog.csdn.net/dba_huangzj/article/details/7699166
------解决方案--------------------
- SQL code
--如果你的系统是2008的，那么使用merge将是完美方案IF EXISTS(SELECT * FROM [dbo].[T_FillingRec1] WHERE SerialPort = @SerialPort And Sequence = @Sequence)BEGIN UPDATE [dbo].[T_FillingRec1] SET [Organize] = @Organize, [BUTime] = @BUTime, [BUUserName] = @BUUserName WHERE SerialPort = @SerialPort And Sequence = @SequenceENDELSEBEGIN Set @FillingRec = NEWID() INSERT INTO [dbo].[T_FillingRec1] ([FillingRec],[Organize],[SerialPort],[Sequence],[BUTime],[BUUserName]) Select @FillingRec,@Organize,@SerialPort,@Sequence,@BUTime,@BUUserNameENDMERGE dbo.[T_FillingRec1] AS TargetUSING (SELECT SerialPort, Sequence,[BUTime],[BUUserName],[Organize] from xxx /*所有的新值*/) AS SourceON (Target.SerialPort = Source.SerialPort AND Target.Sequence = Source.Sequence)WHEN MATCHED THEN UPDATE SET Target.[BUTime] = Source.[BUTime], Target.[BUUserName] = Source.[BUUserName], Target.[Organize] = Source.[Organize],WHEN NOT MATCHED BY TARGET THEN INSERT ([FillingRec],[Organize],[SerialPort],[Sequence],[BUTime],[BUUserName]) VALUES (Source.[FillingRec], Source.[Organize], Source.[SerialPort], Source.[Sequence], Source.[BUTime], Source.[BUUserName]);
------解决方案--------------------
- SQL code
--只是下面这段MERGE dbo.[T_FillingRec1] AS TargetUSING (SELECT SerialPort, Sequence,[BUTime],[BUUserName],[Organize] from xxx /*所有的新值*/) AS SourceON (Target.SerialPort = Source.SerialPort AND Target.Sequence = Source.Sequence)WHEN MATCHED THEN UPDATE SET Target.[BUTime] = Source.[BUTime], Target.[BUUserName] = Source.[BUUserName], Target.[Organize] = Source.[Organize],WHEN NOT MATCHED BY TARGET THEN INSERT ([FillingRec],[Organize],[SerialPort],[Sequence],[BUTime],[BUUserName]) VALUES (Source.[FillingRec], Source.[Organize], Source.[SerialPort], Source.[Sequence], Source.[BUTime], Source.[BUUserName]);