Abstract:As the Internet has entered the era of big data and cloud computing, distributed storage technology has received widespread attention in industry and academia in recent years. In order to solve the problems of excessive server pressure, rapid disaster recovery in remote locations and users' proximity access, this paper analyzes the characteristics, application scenarios and technical challenges of distributed storage based on the theory of CAP and “Multi-active across data centers”, and then uses the open source Redis cache、 RabbitMQ Message Queuing to setup a "Multi-active across data centers" distributed storage system that satisfies the ultimate consistency, availability and partition tolerance. Through the stress test results of the system, it can basically meet the throughput and consistency requirements of industrial applications.