Communications and information are the basis of smart city development. Hence, the performance of information transmissions is vital for the popularity of smart city services. This article is a tutorial study on efficiently managing communication performance in a real-time fashion for smart cities that have a vast number of networking nodes and services. We focus on both QoS and QoE performance and discuss the challenges in guaranteeing such performance by analyzing the characteristics of smart city communication networks. We then explore the opportunity of employing data to automatically manage QoS/ QoE performance by designing a data-driven QoS and QoE framework. This framework implements the functions of training performance prediction and accordingly timely management of QoS/QoE performance via the designed offline and online modules, respectively. We hope that this study may provide preliminary knowledge and research insights for further studies on proactively guaranteeing the performance of a varied number of coexisting and correlated smart city communicators.