I have Modification, Retrieval and Administration operations that are implemented by WCF services. I would like to separate them. The first thing that came to mind was to have three interfaces IRetrieval, IAdministration and IModification and create endpoints based on these interfaces. However, another developer said something about using different bindings or ports. I don't think that's possible - my understanding that WCF binding only defines how the data is treated over the wire and is not fit for logical separation. Am I in the ball park? Are there any other ways to separate the functionality besides the interfaces?
There are only two ways to separate functionalities: separate contracts implemented on the same service or separate contracts implemented on separate services. The difference between these two are more like logical and physical separation. For example default WSDL generation exposes metadata from all implemented contracts in a service. So if you implement all contracts in the single service and expose a metadata endpoint, each client will know an exact description of your administration methods and what security is used.