A client-specific or client-independent polling interval is provided to each client dynamically based on server load. The polling interval can be provided in the server polling response sent in response to a client polling request. The polling interval can be determined using a feedback control system or using a Bucket Reservation Method. The server uses a next polling interval and a flag that indicates if the previous polling request was ignored. Using these two parameters the server can continuously control the polling frequency from the client to achieve optimal performance.