The RGB-D (Red, Green, Blue plus Depth) camera is a widely used onboard sensor for Simultaneous Localization and Mapping (SLAM), as it captures RGB images and depth information simultaneously. This capability has made RGB-D SLAM a topic of broad research interest. We present a comprehensive introduction to the framework of RGB-D SLAM and illustrate representative approaches in each component. Furthermore, we examine the conflicts among diverse performance metrics and discuss performance trade-offs in cutting-edge RGB-D SLAM from a requirement perspective, as achieving a SLAM system with perfect performance is challenging in the present landscape. This manuscript merges theory and practice, serving as a reference for the selection of RGB-D SLAM methods under various practical applications and offering a structured perspective for future development of RGB-D SLAM.