Abstract:Distributed cache is widely used to solve the performance bottleneck problem in traditional relational database,but when third-party applications that are not cache-aware update the back-end database,the distributed cache will end up in an inconsistent state,which has the problem of stale cache data.This paper proposes a distributed cache consistency strategy based on change data capture mechanism.The work integrates trigger-based and log-based change data capture mechanism that can get the real-time data from backend database,and implements data model transformation and SQL translation engine,which can update cache in real-time to guarantee distributed cache coherence.The experiment simulates the key operation in TPC-W benchmark,which verifies that the change data capture based on log has the better database performance and cache consistency effects compared with the change data capture based on trigger.