For small static concentrators, where embedded devices should be individually provisioned, there's strictly speaking not much more to do other than to manage embedded node references in your code when you interact with the Thing Registry, and when you manage sensor data and control requests from clients. The Thing Registry manages the discovery process and provisioning is transparent to the device. Your SensorServer and ControlServer instances manage requests for embedded nodes.

But this method does not allow clients, such as the owner of a concentrator, to browse available nodes on the device, and certainly not if the embedded nodes are not individually provisioned, or if more administrative tasks are required. The ConcentratorServer class can help you to build a more dynamic concentrator that allows clients with sufficient access privileges to administer the concentrator. It also creates an object model, which makes it easier for developers to define data sources and node structures.