As an emerging cybersecurity concept， Zero Trust （ZT） aims at changing the traditional “trust but verify” mode to achieve stricter access control and security defense in networks. The ZT architectures are based on the fundamental premise of not trusting any user， device， or network， and requiring user identity verification and fine-grained authorization control for every access request. First， an elaboration on the concept and core principles of ZT was carried out. Then， the main application scenarios and components of the ZT architectures were introduced， as well as the working principles of these architectures. At the same time， the existing ZT architectures were compared and analyzed， for understanding the specific implementation methods of the ZT concept and providing valuable references for constructing ZT architectures. Finally， the limitations of the ZT architectures were summarized， thereby providing directions for optimizing the application of ZT.