A real-time data storage and retrieval mechanism with the management of multi-replication based on distributed architecture was proposed to solve the deficiencies in throughput, fault tolerance, and scalability of traditional centralized real-time database. Combining the consistent hashing algorithm with data organization based on multi-version control, the mechanism was designed and optimized from three aspects of realtime data storage, management and query. It implements backup synchronization and consistency repair under the premise of ensuring real-time data access, thus improves mass data storage and processing capacity of realtime database, and lays a good foundation on the changes of realtime database distributed storage, multiple redundant backup, dynamically adjustment of system’s scale and other directions. By simulation and testing the mechanism in the prototype system, it is shown that this mechanism can achieve the design targets and performance requirements.