The National Early Warning Release System (NEWRS) is an essential part of the National Emergency Command and Dispatch Platform System. It serves as a key transmission hub connecting various departments and plays a very important role in disseminating warning information. With the increase in the variety and volume of warning data, the NEWRS faces challenges. The capacity and throughput of the NEWRS are under intense restrictions. Problems like low transmission efficiency, limited data processing capability, and high system failure are serious. These problems affect the timely availability of disseminating warning information. The phenomenon that a large amount of data can only exist in the original system is significant. These data are not shared, which does not help to integrate data. It also limits data mining and advanced warning decision-making services. The paper presents a distributed transmission plan for warning data. It promotes a warning transmission model to solve these problems. The transmission model can serve four levels of application: national, provincial, municipal, and county levels. However, it is implemented at national and provincial levels to relieve the instability of the NEWRS caused by multi-level deployment. Meanwhile, it enhances data transmission with caching services using message-oriented middleware technologies. Provincial nodes can serve as a caching function. Once the fault is fixed, provincial nodes can resume transmission automatically. Besides, the transmission model employs distributed clustering technologies to help increase the scalability of the system and reduce the possibility of large-scale failures. It can continue to work when there are some node errors. The warning transmission model improves message-oriented middleware by designing a message data structure, so the transmission model can support dynamic expansion for data types. At the same time, security authentication and data consistency verification for data exchange have been strengthened. All these functions are provided as interface services to facilitate integration with other business systems. To verify these capabilities of the warning transmission model, we conduct experiments in data carrying capacity. Writing a Common Alert Protocol message data requires around 6.7 milliseconds, and reading it takes about 11.4 milliseconds. In timely recovery response, the experiments also show the model is effective. Notably, the transmission model achieves remarkable results in the pilot application in Guizhou Province. The results show that the transmission efficiency of the model is confirmed to be about 94.5 times higher compared to the NEWRS. It has been shown as a practical solution with the potential for nationwide implementation. This will improve the efficiency of communication between national and provincial levels.