A Distributed Disaster Tolerance System Based On RS Code

At present, with the fast development and the fierce competition of the society, amountof data is increasing at an exponential speed. As for the911events in USA, lots of companieshave lost their critical data and bankrupt. There are also other disaster occurs everyday. Forexample, human error, natural environment, network virus infection and so on. What’s more,if the core data lost or damaged or even could not be restored, it is pretty harm for the society.Therefore, it is necessary for us to establish a comprehensive disaster tolerance systemimmediately, which is very important to enterprises and even the country’s prosperity andstability.Erasure codes technology is a kind of recovery technique as well as a pretty importantpart in the disaster tolerance system, which is applied to the local disk data storage. However,with the development of remote disaster tolerance system, it needs an advance erasure codesin order to fit for new tolerance system. Currently, there are a lot of erasure codes used in thedisaster recovery system. Actually, RS code is a higher reliability and suitable cord for massstorage, but the drawback of the low efficieny decoding is obvious.This paper improves original RS code and proposes a new decoding strategy in the lightof the low efficieny decoding, which is more suitable for remote disaster tolerance system. Inadditional, by drawing the characteristics of shorten code and the practical problems, theauthor obtain a reasonable length of data and optimized iterative process, whichcomprehensively improves the fault tolerance of the RS code. Meanwhile, according to afault-tolerant performance evaluation mechanism based on minimum delete list, this paperalso analyzes the tolerance ability of RS code. As a result, the experiments show that theimproved RS code decoding is more efficiently compared to similar erasure codes when thefiles size below50K.What’s more, there is no uniform standards and approaches to disaster systems onevaluation and optimization when it comes to disaster recovery systems now. However, tosolve this problem, author establishes a corresponding mathematical model to quantify theperformance parameters of a disaster recovery system in detail. Eventually, a reasonabledistributed disaster recovery system is verified by experiments for further research workproviding a new idea.