Say I had a large number of objects (e.g. 1000) that I wanted to make available through remoting.
Two plausible architectures are (1) expose each of the objects as a MarshalByRefObject, or (2) register the objects with a broker, and expose only the broker as an MBR object. Aside from the obvious extra level of indirection, how should I evaluate these two approaches? What I'm really wondering is how MarshalByRefObjects are implemented. Does the remoting subsystem maintain a broker of its own for MBR objects? If those 1000 objects were going to exist in memory anyway, what extra overhead is incurred by making them all MBR? Which architecture would be faster? Any comments appreciated. Mike. You can read messages from the Advanced DOTNET archive, unsubscribe from Advanced DOTNET, or subscribe to other DevelopMentor lists at http://discuss.develop.com.
