Smart microgrids are significant in promoting clean energy development and improving microgrid security and reliability. However, harsh environments make them exposed to various hazards, including natural hazards such as hail and wildfire and digital hazards such as cyberattacks. Due to these complex challenges, performing performance evaluation and resilience analysis for smart microgrids in different periods (e.g., before, during, and after the hazards) and different layers (e.g., a data layer and a physical layer) is difficult. To reduce this research gap, this paper develops a new multi-layer failure and multi-dimension resilience methodology in the Internet of Things (IoT). A multi-layer failure mechanism of smart microgrids in IoT with the synergy of the “physical layer, perception layer, communication layer, and application layer” is analysed. A multi-stage performance model for smart microgrids based on operation loops is established. A multi-dimension resilience methodology for smart microgrids is developed with consideration of four performance evolution processes including prevention, degradation, restoration, and reconstitution. At last, a case of Shandong province in China is used to demonstrate the proposed method under normal operating conditions and different types of disasters.