Distributed Data Register High Availability Research And Design Of DOA

Big Data make the Information Technology age come to Data Technology era, the data storage and management has become the core issue. Data Oriented architecture-DOA(Data Oriented Architect) emerges as the times require. When traditional software design facing the challenge of big data, DOA provides a good solution for it. Data Register Controller is the core component of DOA, it register for many type of data and generalized Data, then it forms a logical data resources pool,then the application can access to data conveniently.The high frequency data access and high speed data transmission make the data register controller face a very serious challenge.For the High Availability of distributed systems, there are a variety of solutions. Their core idea is to use backup machine to prevent the failure of NameNode.When it comes, the standby can offer customers by read services. However, these solutions require manual intervention, and also face the possibility of data loss.Therefore, a new idea is appealed to solve the problem of single point of failure in distributed systems.This article proposes a double center server data register controller based onthe consistency of the electionsystem.It can effectively meet the high availability of the system. This article puts forward an algorithm based on a Raft of machine more,it solves the problem of elections and the data sync.In view of how the data register controllerto request distribution and restore downtime, This article proposes a compressors achieve double center server data registration, this method should solve the single point of obstacles and more data consistency problem of machine condition effective.Specifically, the main work is as follows:(1)This articleanalyzed the factors influencing the high availability of the distributed system from reliability and maintainability.According to this train of thought, it researches the current more mature solutions of high availability, then by comparing their advantages and disadvantages, it puts forward distributed data register controller high availability solution.(2) It researches the current distributed consistency algorithm, then choosing the Raft to the consistency algorithm, using the Raft consistency algorithm solves the threeNameNodeleader problems, designed a data register controller data consistency synchronization method, it solve the multiple data consistency problem of machine condition effectively.(3) In order to satisfy the high availability index of the register data controller, this articledesigns a kind of double center server distributed data register controllermodel, it also designs related data structure and interface for this model, and finally transformed the single NameNode communication protocol to apply to three NameNode.(4) Based on election scheme and high availability of the test, then tested the double center server data registercontroller’s feasibility and effectiveness of high availability solution.This article provides a new idea and reference method for distributed data register controller’s High Availability, at the same time, it bases on the election of multi-machine Raft algorithm and data consistency synchronization method that has certain theoretical and practical significance.The main achievements and innovations are as follows:(1)Proposed a multi-machineconsistency of the algorithm based on Raft. Raft algorithm can solve the problem of multi-machineelections, by designing a data register controller of data consistency synchronization method, solved the multiple data consistency problem of machine condition effectively.(2)Proposed a kind of compressors achieve double center server data register controller method. Double center server data register controller solves the bottleneck of the single point of failure, after the election the Leader of registry data consistency of distributed data synchronization method is used to make each is consistent with the data on the NameNode.