Research Of Data Consistency In RAID System

In architecture, the RAID system uses multi‐level, multi‐module design, in which asynchronous mechanism is adopted between modules, and queues are used to exchange data. In technology, redundant technology is brought in to improve the performance of the system and reliability of disk array, but which also brings data consistency problems to the modules in the system. To ensure that the RAID system is running correctly and reliably, the problems of data consistency must be solved.Data consistency is a main line to be researched in distributed and the database fields, in which a lot of special research and theory is proposed. But in the fields of RAID, the study of the problem is relatively weak. According to the existing data consistency theory, the redundancy is the main reason for the problems of data consistency in RAID system, which mainly exists in three forms: the consistency of data between disk and disk, cache and disk, controller and controller.Through the study of RAID technology, and research on system reliability, I propose a series of technology to solve the three aspects problem of data consistency in the system, including: disk data consistency check, power‐fail protection and cache mirror technology and so on.Then I analyze the existing RAID system, and select several data consistency technology according to the RAID system requirements. And then, transform the technology into the design and implementation. In the system, I achieve the following functions: configuration file management system, cache destage system and data consistency check system. And furthermore, I make some improvements to the cache destage system.At last, I make a test to every design. And the results show that the function, to some extent, solve the data consistency problems in the RAID system.