使用開始提交回退避免併發返回語句和UPDLOCK，HOLDLOCK

Question

需要關於下面的存儲過程的建議以避免併發問題。

Create PROCEDURE [SD_SaveUpdate_DepartmentDetails]       
@DeptID int, 
@DepartmentName varchar(100), 
@DepartmentID int out 

AS        
BEGIN 
Begin Try 

-------------------------Insert Update Department Details-----------------------  

Begin 

SET NOCOUNT ON 

BEGIN TRAN Department 

IF EXISTS(SELECT * FROM DepartmentTable WITH (UPDLOCK, HOLDLOCK) WHERE DepartmentName= @DepartmentName) 
BEGIN 
     UPDATE DepartmentTable SET DepartmentName= @DepartmentName WHERE DeptID = @DeptID 
     set @DepartmentID=@DeptID 
END 
ELSE 
BEGIN 
     INSERT INTO DepartmentTable (DepartmentName) VALUES (@DepartmentName) 
     set @DepartmentID=@@IDentity 
END 

COMMIT Transaction Department 
End 
-------------------------Insert Update Department Details-----------------------  
End Try 
Begin Catch 
set @DepartmentID=0 
End Catch 

End 

是我上面的方法使用（UPDLOCK，HOLDLOCK）正確。

我想設置@DepartmentID輸出參數=「-1」，如果部門沒有爲USER2保存，因爲它是重複的。我需要在上面的SP中添加set @ DepartmentID = -1行。

在catch塊我設置@ DepartmentID = 0，如果有一些錯誤，同時保存。但我也想設置@ DepartmentID = -1時重複occcurs。

在博客的其中一篇博文中，我看到使用HOLDLOCK以避免併發症的使用。我想知道其他用戶是否可以在SP運行時訪問（選擇）數據。

你可以請這個SP以及optimitic幫助我（成本效益），以便我可以在所有存儲過程中使用相同的邏輯。

謝謝，

Prathap。

Answer 1

謝謝大家使用鎖定概念time.Without（SET事務隔離級別）或任何UPDLOCK，我想下面的SP

Create PROCEDURE [SD_SaveUpdate_DepartmentDetails]       
@DeptID int, 
@DepartmentName varchar(100), 
@DepartmentID int out 


AS        
BEGIN 
Begin Try 

-------------------------Insert Update Department Details-----------------------  

Begin 

SET NOCOUNT ON 

BEGIN TRAN Department 

IF Not EXISTS(SELECT Top 1 DepartmentName FROM DepartmentTable WHERE DepartmentName= @DepartmentName 
    and Isdeleted=0 and DeptID <> @DeptID) 
and @DeptID > 0 
BEGIN 
     UPDATE DepartmentTable SET DepartmentName= @DepartmentName,lastupdated=getdate() WHERE DeptID = @DeptID 
     set @DepartmentID = @DeptID 
END 
ELSE IF NOT EXISTS(SELECT Top 1 DepartmentName FROM DepartmentTable WHERE DepartmentName= @DepartmentName and IsDeleted=0) 
and @DeptID=0 
BEGIN 
     INSERT INTO DepartmentTable (DepartmentName,isdeleted,lastupdated) VALUES (@DepartmentName,0,getdate()) 
     set @DepartmentID=Scope_IDentity() 
END 
Else 
Begin 
set @DepartmentID = -1 
End 

COMMIT Transaction Department 
End 
-------------------------Insert Update Department Details-----------------------  
End Try 
Begin Catch 
set @DepartmentID = 0 
End Catch 
End 

要確認是否有重複的值，我使用SQLQUERYSTRESS工具爲多用戶場景（併發）及其成功測試了此SP。沒有重複的值。

如果有任何建議，請讓我知道。

Answer 2

你爲什麼要實現所有這些複雜的併發性的東西？

而且從不使用@@IDENTITY，總是使用scope_identity()

我想設置@DepartmentID輸出paramater =「1」，如果該部門未保存爲USER2，因爲它的一個duplicate.Where做我需要在上面的SP中添加set @ DepartmentID = -1行。

做什麼類似下面的問題 - 爲什麼你需要鎖定表？

-- This will insert a record if it doesn't already exist 
-- If it does already exist it won't insert a record 
INSERT INTO DepartmentTable (DeptID) 
SELECT DeptID WHERE NOT EXISTS (
    SELECT 1 FROM DepartmentTable WHERE DeptID = @DeptID 

-- find the new ID. This is NULL if no record was inserted 
SELECT @DepartmentID=SCOPE_IDENTITY() 

-- If no row was inserted, this was a duplicate and scope_identity is null 
IF @DepartmentID IS NULL SET @DepartmentID = -1 

Answer 3

你看了Merge聲明嗎？我認爲Merge會是一個更好的方法。爲了避免對departmentname

CREATE PROCEDURE [SD_SaveUpdate_DepartmentDetails]       
@depatid int, 
@DepartmentName varchar(100) 
AS        
BEGIN 
Begin Try 

-------------------------Insert Update Department Details-----------------------  

Begin  
SET NOCOUNT ON 
BEGIN TRAN Department  

MERGE DepartmentTable TGT 
USING 
(SELECT @departmentname deptname, @departmentid deptid) as SRC 
ON TGT.DEPTID=SRC.DEPTID 
WHEN MATCHED THEN 
    UPDATE 
    SET DepartmentName= @DepartmentName,lastupdated=getdate() 
WHEN NOT MATCHED THEN 
    INSERT VALUES(@DepartmentName,0,getdate()) 

OUTPUT case when $action='INSERT' then DELETED.DEPTID ELSE Inserted.deptid end