The existence of smart cities and the diversity of their applications has now become a necessity that helps in reducing costs and managing available resources efficiently. IoT applications help to expand and support smart cities, as it manages and monitors devices and acquire rapidly a collection of big data to benefit from it. Smart cities applications vary in terms of the types of data generated, the need for quickly data analysis, response speed, or the need for massive data storage. Fog and mist computing models provide solutions for all these requirements, as they can analyze and store big data and interact with IoT devices quickly and smoothly. In this research a fault tolerant model will be proposed to support and manage applications in smart cities. This model relies on three tiers of computing resources (Mist, Fog, and Cloud). The proposed model distributes processing tasks to the edge to reduce data latency and support real-time applications in smart cities. In addition to, the IoT tasks are classified in this research based on their deadline and the urgency of their execution. Also, the performance of the proposed model is measured and compared with three scheduling models, namely; Min-Min, Credit-Based-Scheduling (CBS) and Earliest-Feasible-Deadline-First (EFDF). Through comprehensive simulations, the test measurements obtain improvement in the performance metrics.