Research on Massive Data Oriented Data Consistency
面向海量数据的数据一致性研究

Replication is the key technology of massive data management. Data consistency of replicas is one of the keys for achieving fault tolerance and performance enhancement in distributed systems. Strong consistency ensures that cow current updates will not conflict but limits system availability, throughput, and the practical degree of replication, while weak consistency only guarantees eventual agreement but provides strong fault tolerance. The current techniques of research in data consistency are studied. Then, according to the characteristic of massive data, the problems such as update issuer, update propagation manner, update propagation content and conflict resolution that should be studied in data consistency are brought forward and the feasible methods are proposed respectively.