I'm using the following code in a singleton remoting object MBRO. This function is only called on the server side.
''' <summary>
''' Return a cached DataCentricObject
''' </summary>
''' <created_by>CodingBarfield</created_by>
''' <date>04-08-2011</date>
Function DataCentricObjectName(ByVal intID As Integer) As String
Try
SyncLock dictDataCentricObject
If Not dictDataCentricObject.ContainsKey(intID) Then
Dim st As struct = dcLoader.LoadRecord(intID)
dictDataCentricObject(intID) = st.Descript
End If
Return dictDataCentricObject(intID)
End SyncLock
Catch ex As Exception
Throw New Exception("Error in GetTargName", ex)
End Try
End Function
Private dictDataCentricObject As New Dictionary(Of Integer, String)
Dim dcLoader As New DataCentricObject
The LoadRecord function simply reads a line from a database table and copies the fields into 开发者_Python百科a little data structure.
The question
- Is this code multithreading safe (in a remoting environment)
- Are there any performance benefits for different code
It depends on what dcLoader.LoadRecord
does. I'm going to take a guess that this just reads some data and does not update any state, I'm also assuming that other accesors of dictDataCentricObject
lock on that object. If this is true then I think this code is thread-safe.
If dcLoader.LoadRecord
was inexpensive you could do it before your lock to improve concurrency. However, I suspect it results in a call to the database so for overall performance it may be better to stay where it is. This does mean that calls to the function will serialise on access to a more expensive resource. If, and only if, this causes you a performance problem you could implement some caching around dcLoader.LoadRecord
.
精彩评论