Research On Distributed Data Consistency Method Based On Rainbow Consistency

Since the beginning of the 21 st century,large-scale distributed systems and cloud computing have become increasingly popular,the application value and importance of distributed technologies have become increasingly prominent.In order to meet the requirements of the new platform and new computing model for system performance,availability,fault tolerance,scalability and other indicators,the underlying data storage system usually uses distributed data technology,which includes data partitioning technology and data replication technology.However,distributed data brings data consistency problems.Data consistency is a key issue to ensure application data security and efficiency.It is especially important in the distributed field.But,according to the CAP theorem,there is a mutual constraint between data consistency and system high availability and partition fault tolerance,the two can coexist at the same time.In the distributed system,the fault tolerance of the partition must be guaranteed,resulting in consistency and latency(the manifestation of system availability)is difficult to reconcile.Therefore,data consistency issues have become a challenging research topic in distributed data services.In view of the above problems,aside from the previous research perspective,that various consistency models are mixed and improved in various ways,turn to the perspective of operation.Taking the idea of continuous fine-grained partitioning of the consistency of operations required,a new data consistency model,rainbow consistency,is proposed.Based on the model,the consistency problem of distributed data is optimized,and the performance of the distributed system is improved.The core idea of the model is that the consistency of the operational consistency can be dynamically divided according to the specific application,so that the consistency effect of the different levels of the division is optimal,that is,the delay is reduced as much as possible to ensure consistency,with adjustability.Combining the characteristics of system state convergence and conventional quantity security,this model proposes a marking method to classify different strong and weak consistency operations and make corresponding rainbow color markers to distinguish the same nature of operations,Avoid mixing operations with different consistency requirements and causing poor system performance.Based on the rainbow consistency model,the RbS distributed framework is designed.The token control mechanism is adopted to adjust the operation consistency level selection of different color markers,so that different operations can meet different consistency requirements and the pertinence is strong.On this basis,through benchmark experiments,comparative experiments and a specific example,the validity,superiority and robustness of the consistency model are improved by comparing the user response time delay and system throughput,and improve the user experience.Experiments show that the rainbow consistency model has certain practical guiding significance for the consistency requirement classification of different applications.