Fault Tolerance Technique Based on State Real-time Synchronization
Zhenpeng Xu, Hongwei Liu, Yu Liu · 2021 IEEE International Conference on Data Science and Computer Application (ICDSCA) · 2021
Key service hosts in information systems have higher and higher requirements on high reliability and survivability. Combined with the development trend of fault-tolerant technology, the mechanisms of data synchronization, memory synchronization, IO synchronization, fault detection, and transparent switching between active and standby are introduced, and a dual-machine redundancy high availability model based on state real-time synchronization is constructed. In order to reduce the overhead of memory state synchronization, an event-driven memory synchronization trigger mechanism is designed, that is, the key IO event-driven event trigger mechanism is adopted for the synchronization timing of memory checkpoints, so as to minimize unnecessary memory data update operations between the main and standby computers and reduce the overhead of computer state synchronization. At the same time, combined with BMC monitoring interface based on physical host, a fine-grained fault detection method is proposed. Related work can greatly reduce the overhead caused by high availability mechanism, and reduce the impact of fault-tolerant mechanism on system computing performance to a certain extent. At the same time, it has the ability to switch between active and standby quickly, which is helpful to better guarantee the uninterrupted and continuous running ability of task applications in key systems.