How We Reduced Multi-Region Read Latency and Network Traffic by 50%

In data-critical industries like the financial industry, data storage and access need a high availability architecture. Within this architecture, if a server goes down, the system can still provide services. But if a natural disaster like an earthquake occurs, there might be a network partition or, even worse, all the servers in the data center (DC) could fail. In this case, to guarantee system availability, we need a remote disaster recovery architecture. What's more, it's better to keep a long-distance between data centers, so that a disaster in one region won't make a huge impact on another.

However, this architecture has two limitations: