Towards Processing Capacity Prediction for the Cloud: Theoretical Models