For big data, it is important to effectively store and access the exponentially growing volume of data. Distributed databases connect databases on various machines through the network, forming a larger and more complex database. However, its drawbacks lead to performance bottlenecks, such as unreliable networks and situations where the same key has multiple versions of values in the distributed database system. This confuses the way of data access. Therefore, we integrate emerging memory into the storage architecture, leveraging its advantages to reduce the performance costs caused by data inconsistency in distributed databases. The environment is adjusted to better integrate emerging memory [10]. The tiered storage mechanism to a three-tier storage architecture reduces reading time by 57%.